explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-12-24 00:46:28 +03:00
Code Issues Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'origin/develop' into feature/better-parser

Browse Source
This commit is contained in:
Matthew Honnibal 2017-10-27 12:34:48 +00:00
commit 531142a933
78 changed files with 1655 additions and 3278 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.rst
View File

@ -42,7 +42,7 @@ integration. It's commercial open-source software, released under the MIT licens
=================== ===
`spaCy 101`_ New to spaCy? Here's everything you need to know!
`Usage Guides`_ How to use spaCy and its features.
`New in v2.0`_ New features, backwards incompatibilitiies and migration guide.
`New in v2.0`_ New features, backwards incompatibilities and migration guide.
`API Reference`_ The detailed reference for spaCy's API.
`Models`_ Download statistical language models for spaCy.
`Resources`_ Libraries, extensions, demos, books and courses.

22
examples/README.md
View File

@ -2,20 +2,18 @@
# spaCy examples
The examples are Python scripts with well-behaved command line interfaces. For a full list of spaCy tutorials and code snippets, see the [documentation](https://spacy.io/docs/usage/tutorials).
The examples are Python scripts with well-behaved command line interfaces. For
more detailed usage guides, see the [documentation](https://alpha.spacy.io/usage/).
## How to run an example
For example, to run the [`nn_text_class.py`](nn_text_class.py) script, do:
To see the available arguments, you can use the `--help` or `-h` flag:
```bash
$ python examples/nn_text_class.py
usage: nn_text_class.py [-h] [-d 3] [-H 300] [-i 5] [-w 40000] [-b 24]
[-r 0.3] [-p 1e-05] [-e 0.005]
data_dir
nn_text_class.py: error: too few arguments
$ python examples/training/train_ner.py --help
```
You can print detailed help with the `-h` argument.
While we try to keep the examples up to date, they are not currently exercised by the test suite, as some of them require significant data downloads or take time to train. If you find that an example is no longer running, [please tell us](https://github.com/explosion/spaCy/issues)! We know there's nothing worse than trying to figure out what you're doing wrong, and it turns out your code was never the problem.
While we try to keep the examples up to date, they are not currently exercised
by the test suite, as some of them require significant data downloads or take
time to train. If you find that an example is no longer running,
[please tell us](https://github.com/explosion/spaCy/issues)! We know there's
nothing worse than trying to figure out what you're doing wrong, and it turns
out your code was never the problem.

37
examples/_handler.py
View File

@ -1,37 +0,0 @@
# encoding: utf8
from __future__ import unicode_literals, print_function
from math import sqrt
from numpy import dot
from numpy.linalg import norm
def handle_tweet(spacy, tweet_data, query):
text = tweet_data.get('text', u'')
# Twython returns either bytes or unicode, depending on tweet.
# ಠ_ಠ #APIshaming
try:
match_tweet(spacy, text, query)
except TypeError:
match_tweet(spacy, text.decode('utf8'), query)
def match_tweet(spacy, text, query):
def get_vector(word):
return spacy.vocab[word].repvec
tweet = spacy(text)
tweet = [w.repvec for w in tweet if w.is_alpha and w.lower_ != query]
if tweet:
accept = map(get_vector, 'child classroom teach'.split())
reject = map(get_vector, 'mouth hands giveaway'.split())
y = sum(max(cos(w1, w2), 0) for w1 in tweet for w2 in accept)
n = sum(max(cos(w1, w2), 0) for w1 in tweet for w2 in reject)
if (y / (y + n)) >= 0.5 or True:
print(text)
def cos(v1, v2):
return dot(v1, v2) / (norm(v1) * norm(v2))

59
examples/get_parse_subregions.py
View File

@ -1,59 +0,0 @@
"""Issue #252
Question:
In the documents and tutorials the main thing I haven't found is examples on how to break sentences down into small sub thoughts/chunks. The noun_chunks is handy, but having examples on using the token.head to find small (near-complete) sentence chunks would be neat.
Lets take the example sentence on https://displacy.spacy.io/displacy/index.html
displaCy uses CSS and JavaScript to show you how computers understand language
This sentence has two main parts (XCOMP & CCOMP) according to the breakdown:
[displaCy] uses CSS and Javascript [to + show]
&
show you how computers understand [language]
I'm assuming that we can use the token.head to build these groups. In one of your examples you had the following function.
def dependency_labels_to_root(token):
'''Walk up the syntactic tree, collecting the arc labels.'''
dep_labels = []
while token.head is not token:
dep_labels.append(token.dep)
token = token.head
return dep_labels
"""
from __future__ import print_function, unicode_literals
# Answer:
# The easiest way is to find the head of the subtree you want, and then use the
# `.subtree`, `.children`, `.lefts` and `.rights` iterators. `.subtree` is the
# one that does what you're asking for most directly:
from spacy.en import English
nlp = English()
doc = nlp(u'displaCy uses CSS and JavaScript to show you how computers understand language')
for word in doc:
if word.dep_ in ('xcomp', 'ccomp'):
print(''.join(w.text_with_ws for w in word.subtree))
# It'd probably be better for `word.subtree` to return a `Span` object instead
# of a generator over the tokens. If you want the `Span` you can get it via the
# `.right_edge` and `.left_edge` properties. The `Span` object is nice because
# you can easily get a vector, merge it, etc.
doc = nlp(u'displaCy uses CSS and JavaScript to show you how computers understand language')
for word in doc:
if word.dep_ in ('xcomp', 'ccomp'):
subtree_span = doc[word.left_edge.i : word.right_edge.i + 1]
print(subtree_span.text, '|', subtree_span.root.text)
print(subtree_span.similarity(doc))
print(subtree_span.similarity(subtree_span.root))
# You might also want to select a head, and then select a start and end position by
# walking along its children. You could then take the `.left_edge` and `.right_edge`
# of those tokens, and use it to calculate a span.

59
examples/information_extraction.py
View File

@ -1,59 +0,0 @@
import plac
from spacy.en import English
from spacy.parts_of_speech import NOUN
from spacy.parts_of_speech import ADP as PREP
def _span_to_tuple(span):
start = span[0].idx
end = span[-1].idx + len(span[-1])
tag = span.root.tag_
text = span.text
label = span.label_
return (start, end, tag, text, label)
def merge_spans(spans, doc):
# This is a bit awkward atm. What we're doing here is merging the entities,
# so that each only takes up a single token. But an entity is a Span, and
# each Span is a view into the doc. When we merge a span, we invalidate
# the other spans. This will get fixed --- but for now the solution
# is to gather the information first, before merging.
tuples = [_span_to_tuple(span) for span in spans]
for span_tuple in tuples:
doc.merge(*span_tuple)
def extract_currency_relations(doc):
merge_spans(doc.ents, doc)
merge_spans(doc.noun_chunks, doc)
relations = []
for money in filter(lambda w: w.ent_type_ == 'MONEY', doc):
if money.dep_ in ('attr', 'dobj'):
subject = [w for w in money.head.lefts if w.dep_ == 'nsubj']
if subject:
subject = subject[0]
relations.append((subject, money))
elif money.dep_ == 'pobj' and money.head.dep_ == 'prep':
relations.append((money.head.head, money))
return relations
def main():
nlp = English()
texts = [
u'Net income was $9.4 million compared to the prior year of $2.7 million.',
u'Revenue exceeded twelve billion dollars, with a loss of $1b.',
]
for text in texts:
doc = nlp(text)
relations = extract_currency_relations(doc)
for r1, r2 in relations:
print(r1.text, r2.ent_type_, r2.text)
if __name__ == '__main__':
plac.call(main)

62
examples/information_extraction/entity_relations.py Normal file
View File

@ -0,0 +1,62 @@
#!/usr/bin/env python
# coding: utf8
"""
A simple example of extracting relations between phrases and entities using
spaCy's named entity recognizer and the dependency parse. Here, we extract
money and currency values (entities labelled as MONEY) and then check the
dependency tree to find the noun phrase they are referring to – for example:
$9.4 million --> Net income.
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
import spacy
TEXTS = [
'Net income was $9.4 million compared to the prior year of $2.7 million.',
'Revenue exceeded twelve billion dollars, with a loss of $1b.',
]
@plac.annotations(
model=("Model to load (needs parser and NER)", "positional", None, str))
def main(model='en_core_web_sm'):
nlp = spacy.load(model)
print("Loaded model '%s'" % model)
print("Processing %d texts" % len(TEXTS))
for text in TEXTS:
doc = nlp(text)
relations = extract_currency_relations(doc)
for r1, r2 in relations:
print('{:<10}\t{}\t{}'.format(r1.text, r2.ent_type_, r2.text))
def extract_currency_relations(doc):
# merge entities and noun chunks into one token
for span in [*list(doc.ents), *list(doc.noun_chunks)]:
span.merge()
relations = []
for money in filter(lambda w: w.ent_type_ == 'MONEY', doc):
if money.dep_ in ('attr', 'dobj'):
subject = [w for w in money.head.lefts if w.dep_ == 'nsubj']
if subject:
subject = subject[0]
relations.append((subject, money))
elif money.dep_ == 'pobj' and money.head.dep_ == 'prep':
relations.append((money.head.head, money))
return relations
if __name__ == '__main__':
plac.call(main)
# Expected output:
# Net income MONEY $9.4 million
# the prior year MONEY $2.7 million
# Revenue MONEY twelve billion dollars
# a loss MONEY 1b

65
examples/information_extraction/parse_subtrees.py Normal file
View File

@ -0,0 +1,65 @@
#!/usr/bin/env python
# coding: utf8
"""
This example shows how to navigate the parse tree including subtrees attached
to a word.
Based on issue #252:
"In the documents and tutorials the main thing I haven't found is
examples on how to break sentences down into small sub thoughts/chunks. The
noun_chunks is handy, but having examples on using the token.head to find small
(near-complete) sentence chunks would be neat. Lets take the example sentence:
"displaCy uses CSS and JavaScript to show you how computers understand language"
This sentence has two main parts (XCOMP & CCOMP) according to the breakdown:
[displaCy] uses CSS and Javascript [to + show]
show you how computers understand [language]
I'm assuming that we can use the token.head to build these groups."
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
import spacy
@plac.annotations(
model=("Model to load", "positional", None, str))
def main(model='en_core_web_sm'):
nlp = spacy.load(model)
print("Loaded model '%s'" % model)
doc = nlp("displaCy uses CSS and JavaScript to show you how computers "
"understand language")
# The easiest way is to find the head of the subtree you want, and then use
# the `.subtree`, `.children`, `.lefts` and `.rights` iterators. `.subtree`
# is the one that does what you're asking for most directly:
for word in doc:
if word.dep_ in ('xcomp', 'ccomp'):
print(''.join(w.text_with_ws for w in word.subtree))
# It'd probably be better for `word.subtree` to return a `Span` object
# instead of a generator over the tokens. If you want the `Span` you can
# get it via the `.right_edge` and `.left_edge` properties. The `Span`
# object is nice because you can easily get a vector, merge it, etc.
for word in doc:
if word.dep_ in ('xcomp', 'ccomp'):
subtree_span = doc[word.left_edge.i : word.right_edge.i + 1]
print(subtree_span.text, '|', subtree_span.root.text)
# You might also want to select a head, and then select a start and end
# position by walking along its children. You could then take the
# `.left_edge` and `.right_edge` of those tokens, and use it to calculate
# a span.
if __name__ == '__main__':
plac.call(main)
# Expected output:
# to show you how computers understand language
# how computers understand language
# to show you how computers understand language | show
# how computers understand language | understand

60
examples/phrase_matcher.py → examples/information_extraction/phrase_matcher.py
View File

@ -4,22 +4,24 @@ The idea is to associate each word in the vocabulary with a tag, noting whether
they begin, end, or are inside at least one pattern. An additional tag is used
for single-word patterns. Complete patterns are also stored in a hash set.
When we process a document, we look up the words in the vocabulary, to associate
the words with the tags. We then search for tag-sequences that correspond to
valid candidates. Finally, we look up the candidates in the hash set.
When we process a document, we look up the words in the vocabulary, to
associate the words with the tags. We then search for tag-sequences that
correspond to valid candidates. Finally, we look up the candidates in the hash
set.
For instance, to search for the phrases "Barack Hussein Obama" and "Hilary Clinton", we
would associate "Barack" and "Hilary" with the B tag, Hussein with the I tag,
and Obama and Clinton with the L tag.
For instance, to search for the phrases "Barack Hussein Obama" and "Hilary
Clinton", we would associate "Barack" and "Hilary" with the B tag, Hussein with
the I tag, and Obama and Clinton with the L tag.
The document "Barack Clinton and Hilary Clinton" would have the tag sequence
[{B}, {L}, {}, {B}, {L}], so we'd get two matches. However, only the second candidate
is in the phrase dictionary, so only one is returned as a match.
[{B}, {L}, {}, {B}, {L}], so we'd get two matches. However, only the second
candidate is in the phrase dictionary, so only one is returned as a match.
The algorithm is O(n) at run-time for document of length n because we're only ever
matching over the tag patterns. So no matter how many phrases we're looking for,
our pattern set stays very small (exact size depends on the maximum length we're
looking for, as the query language currently has no quantifiers)
The algorithm is O(n) at run-time for document of length n because we're only
ever matching over the tag patterns. So no matter how many phrases we're
looking for, our pattern set stays very small (exact size depends on the
maximum length we're looking for, as the query language currently has no
quantifiers).
The example expects a .bz2 file from the Reddit corpus, and a patterns file,
formatted in jsonl as a sequence of entries like this:
@ -32,11 +34,9 @@ formatted in jsonl as a sequence of entries like this:
{"text":"Argentina"}
"""
from __future__ import print_function, unicode_literals, division
from bz2 import BZ2File
import time
import math
import codecs
import plac
import ujson
@ -44,6 +44,24 @@ from spacy.matcher import PhraseMatcher
import spacy
@plac.annotations(
patterns_loc=("Path to gazetteer", "positional", None, str),
text_loc=("Path to Reddit corpus file", "positional", None, str),
n=("Number of texts to read", "option", "n", int),
lang=("Language class to initialise", "option", "l", str))
def main(patterns_loc, text_loc, n=10000, lang='en'):
nlp = spacy.blank('en')
nlp.vocab.lex_attr_getters = {}
phrases = read_gazetteer(nlp.tokenizer, patterns_loc)
count = 0
t1 = time.time()
for ent_id, text in get_matches(nlp.tokenizer, phrases,
read_text(text_loc, n=n)):
count += 1
t2 = time.time()
print("%d docs in %.3f s. %d matches" % (n, (t2 - t1), count))
def read_gazetteer(tokenizer, loc, n=-1):
for i, line in enumerate(open(loc)):
data = ujson.loads(line.strip())
@ -75,18 +93,6 @@ def get_matches(tokenizer, phrases, texts, max_length=6):
yield (ent_id, doc[start:end].text)
def main(patterns_loc, text_loc, n=10000):
nlp = spacy.blank('en')
nlp.vocab.lex_attr_getters = {}
phrases = read_gazetteer(nlp.tokenizer, patterns_loc)
count = 0
t1 = time.time()
for ent_id, text in get_matches(nlp.tokenizer, phrases, read_text(text_loc, n=n)):
count += 1
t2 = time.time()
print("%d docs in %.3f s. %d matches" % (n, (t2 - t1), count))
if __name__ == '__main__':
if False:
import cProfile

5
examples/inventory_count/Instructions.md
View File

@ -1,5 +0,0 @@
An example of inventory counting using SpaCy.io NLP library. Meant to show how to instantiate Spacy's English class, and allow reusability by reloading the main module.
In the future, a better implementation of this library would be to apply machine learning to each query and learn what to classify as the quantitative statement (55 kgs OF), vs the actual item of count (how likely is a preposition object to be the item of count if x,y,z qualifications appear in the statement).

35
examples/inventory_count/inventory.py
View File

@ -1,35 +0,0 @@
class Inventory:
"""
Inventory class - a struct{} like feature to house inventory counts
across modules.
"""
originalQuery = None
item = ""
unit = ""
amount = ""
def __init__(self, statement):
"""
Constructor - only takes in the original query/statement
:return: new Inventory object
"""
self.originalQuery = statement
pass
def __str__(self):
return str(self.amount) + ' ' + str(self.unit) + ' ' + str(self.item)
def printInfo(self):
print '-------------Inventory Count------------'
print "Original Query: " + str(self.originalQuery)
print 'Amount: ' + str(self.amount)
print 'Unit: ' + str(self.unit)
print 'Item: ' + str(self.item)
print '----------------------------------------'
def isValid(self):
if not self.item or not self.unit or not self.amount or not self.originalQuery:
return False
else:
return True

92
examples/inventory_count/inventoryCount.py
View File

@ -1,92 +0,0 @@
from inventory import Inventory
def runTest(nlp):
testset = []
testset += [nlp(u'6 lobster cakes')]
testset += [nlp(u'6 avacados')]
testset += [nlp(u'fifty five carrots')]
testset += [nlp(u'i have 55 carrots')]
testset += [nlp(u'i got me some 9 cabbages')]
testset += [nlp(u'i got 65 kgs of carrots')]
result = []
for doc in testset:
c = decodeInventoryEntry_level1(doc)
if not c.isValid():
c = decodeInventoryEntry_level2(doc)
result.append(c)
for i in result:
i.printInfo()
def decodeInventoryEntry_level1(document):
"""
Decodes a basic entry such as: '6 lobster cake' or '6' cakes
@param document : NLP Doc object
:return: Status if decoded correctly (true, false), and Inventory object
"""
count = Inventory(str(document))
for token in document:
if token.pos_ == (u'NOUN' or u'NNS' or u'NN'):
item = str(token)
for child in token.children:
if child.dep_ == u'compound' or child.dep_ == u'ad':
item = str(child) + str(item)
elif child.dep_ == u'nummod':
count.amount = str(child).strip()
for numerical_child in child.children:
# this isn't arithmetic rather than treating it such as a string
count.amount = str(numerical_child) + str(count.amount).strip()
else:
print "WARNING: unknown child: " + str(child) + ':'+str(child.dep_)
count.item = item
count.unit = item
return count
def decodeInventoryEntry_level2(document):
"""
Entry level 2, a more complicated parsing scheme that covers examples such as
'i have 80 boxes of freshly baked pies'
@document @param document : NLP Doc object
:return: Status if decoded correctly (true, false), and Inventory object-
"""
count = Inventory(str(document))
for token in document:
# Look for a preposition object that is a noun (this is the item we are counting).
# If found, look at its' dependency (if a preposition that is not indicative of
# inventory location, the dependency of the preposition must be a noun
if token.dep_ == (u'pobj' or u'meta') and token.pos_ == (u'NOUN' or u'NNS' or u'NN'):
item = ''
# Go through all the token's children, these are possible adjectives and other add-ons
# this deals with cases such as 'hollow rounded waffle pancakes"
for i in token.children:
item += ' ' + str(i)
item += ' ' + str(token)
count.item = item
# Get the head of the item:
if token.head.dep_ != u'prep':
# Break out of the loop, this is a confusing entry
break
else:
amountUnit = token.head.head
count.unit = str(amountUnit)
for inner in amountUnit.children:
if inner.pos_ == u'NUM':
count.amount += str(inner)
return count

30
examples/inventory_count/main.py
View File

@ -1,30 +0,0 @@
import inventoryCount as mainModule
import os
from spacy.en import English
if __name__ == '__main__':
"""
Main module for this example - loads the English main NLP class,
and keeps it in RAM while waiting for the user to re-run it. Allows the
developer to re-edit their module under testing without having
to wait as long to load the English class
"""
# Set the NLP object here for the parameters you want to see,
# or just leave it blank and get all the opts
print "Loading English module... this will take a while."
nlp = English()
print "Done loading English module."
while True:
try:
reload(mainModule)
mainModule.runTest(nlp)
raw_input('================ To reload main module, press Enter ================')
except Exception, e:
print "Unexpected error: " + str(e)
continue

161
examples/matcher_example.py
View File

@ -1,161 +0,0 @@
from __future__ import unicode_literals, print_function
import spacy.en
import spacy.matcher
from spacy.attrs import ORTH, TAG, LOWER, IS_ALPHA, FLAG63
import plac
def main():
nlp = spacy.en.English()
example = u"I prefer Siri to Google Now. I'll google now to find out how the google now service works."
before = nlp(example)
print("Before")
for ent in before.ents:
print(ent.text, ent.label_, [w.tag_ for w in ent])
# Output:
# Google ORG [u'NNP']
# google ORG [u'VB']
# google ORG [u'NNP']
nlp.matcher.add(
"GoogleNow", # Entity ID: Not really used at the moment.
"PRODUCT", # Entity type: should be one of the types in the NER data
{"wiki_en": "Google_Now"}, # Arbitrary attributes. Currently unused.
[ # List of patterns that can be Surface Forms of the entity
# This Surface Form matches "Google Now", verbatim
[ # Each Surface Form is a list of Token Specifiers.
{ # This Token Specifier matches tokens whose orth field is "Google"
ORTH: "Google"
},
{ # This Token Specifier matches tokens whose orth field is "Now"
ORTH: "Now"
}
],
[ # This Surface Form matches "google now", verbatim, and requires
# "google" to have the NNP tag. This helps prevent the pattern from
# matching cases like "I will google now to look up the time"
{
ORTH: "google",
TAG: "NNP"
},
{
ORTH: "now"
}
]
]
)
after = nlp(example)
print("After")
for ent in after.ents:
print(ent.text, ent.label_, [w.tag_ for w in ent])
# Output
# Google Now PRODUCT [u'NNP', u'RB']
# google ORG [u'VB']
# google now PRODUCT [u'NNP', u'RB']
#
# You can customize attribute values in the lexicon, and then refer to the
# new attributes in your Token Specifiers.
# This is particularly good for word-set membership.
#
australian_capitals = ['Brisbane', 'Sydney', 'Canberra', 'Melbourne', 'Hobart',
'Darwin', 'Adelaide', 'Perth']
# Internally, the tokenizer immediately maps each token to a pointer to a
# LexemeC struct. These structs hold various features, e.g. the integer IDs
# of the normalized string forms.
# For our purposes, the key attribute is a 64-bit integer, used as a bit field.
# spaCy currently only uses 12 of the bits for its built-in features, so
# the others are available for use. It's best to use the higher bits, as
# future versions of spaCy may add more flags. For instance, we might add
# a built-in IS_MONTH flag, taking up FLAG13. So, we bind our user-field to
# FLAG63 here.
is_australian_capital = FLAG63
# Now we need to set the flag value. It's False on all tokens by default,
# so we just need to set it to True for the tokens we want.
# Here we iterate over the strings, and set it on only the literal matches.
for string in australian_capitals:
lexeme = nlp.vocab[string]
lexeme.set_flag(is_australian_capital, True)
print('Sydney', nlp.vocab[u'Sydney'].check_flag(is_australian_capital))
print('sydney', nlp.vocab[u'sydney'].check_flag(is_australian_capital))
# If we want case-insensitive matching, we have to be a little bit more
# round-about, as there's no case-insensitive index to the vocabulary. So
# we have to iterate over the vocabulary.
# We'll be looking up attribute IDs in this set a lot, so it's good to pre-build it
target_ids = {nlp.vocab.strings[s.lower()] for s in australian_capitals}
for lexeme in nlp.vocab:
if lexeme.lower in target_ids:
lexeme.set_flag(is_australian_capital, True)
print('Sydney', nlp.vocab[u'Sydney'].check_flag(is_australian_capital))
print('sydney', nlp.vocab[u'sydney'].check_flag(is_australian_capital))
print('SYDNEY', nlp.vocab[u'SYDNEY'].check_flag(is_australian_capital))
# Output
# Sydney True
# sydney False
# Sydney True
# sydney True
# SYDNEY True
#
# The key thing to note here is that we're setting these attributes once,
# over the vocabulary --- and then reusing them at run-time. This means the
# amortized complexity of anything we do this way is going to be O(1). You
# can match over expressions that need to have sets with tens of thousands
# of values, e.g. "all the street names in Germany", and you'll still have
# O(1) complexity. Most regular expression algorithms don't scale well to
# this sort of problem.
#
# Now, let's use this in a pattern
nlp.matcher.add("AuCitySportsTeam", "ORG", {},
[
[
{LOWER: "the"},
{is_australian_capital: True},
{TAG: "NNS"}
],
[
{LOWER: "the"},
{is_australian_capital: True},
{TAG: "NNPS"}
],
[
{LOWER: "the"},
{IS_ALPHA: True}, # Allow a word in between, e.g. The Western Sydney
{is_australian_capital: True},
{TAG: "NNS"}
],
[
{LOWER: "the"},
{IS_ALPHA: True}, # Allow a word in between, e.g. The Western Sydney
{is_australian_capital: True},
{TAG: "NNPS"}
]
])
doc = nlp(u'The pattern should match the Brisbane Broncos and the South Darwin Spiders, but not the Colorado Boulders')
for ent in doc.ents:
print(ent.text, ent.label_)
# Output
# the Brisbane Broncos ORG
# the South Darwin Spiders ORG
# Output
# Before
# Google ORG [u'NNP']
# google ORG [u'VB']
# google ORG [u'NNP']
# After
# Google Now PRODUCT [u'NNP', u'RB']
# google ORG [u'VB']
# google now PRODUCT [u'NNP', u'RB']
# Sydney True
# sydney False
# Sydney True
# sydney True
# SYDNEY True
# the Brisbane Broncos ORG
# the South Darwin Spiders ORG
if __name__ == '__main__':
main()

74
examples/parallel_parse.py
View File

@ -1,74 +0,0 @@
from __future__ import print_function, unicode_literals, division
import io
import bz2
import logging
from toolz import partition
from os import path
import re
import spacy.en
from spacy.tokens import Doc
from joblib import Parallel, delayed
import plac
import ujson
def parallelize(func, iterator, n_jobs, extra, backend='multiprocessing'):
extra = tuple(extra)
return Parallel(n_jobs=n_jobs, backend=backend)(delayed(func)(*(item + extra))
for item in iterator)
def iter_comments(loc):
with bz2.BZ2File(loc) as file_:
for i, line in enumerate(file_):
yield ujson.loads(line)['body']
pre_format_re = re.compile(r'^[\`\*\~]')
post_format_re = re.compile(r'[\`\*\~]$')
url_re = re.compile(r'\[([^]]+)\]\(%%URL\)')
link_re = re.compile(r'\[([^]]+)\]\(https?://[^\)]+\)')
def strip_meta(text):
text = link_re.sub(r'\1', text)
text = text.replace('&gt;', '>').replace('&lt;', '<')
text = pre_format_re.sub('', text)
text = post_format_re.sub('', text)
return text.strip()
def save_parses(batch_id, input_, out_dir, n_threads, batch_size):
out_loc = path.join(out_dir, '%d.bin' % batch_id)
if path.exists(out_loc):
return None
print('Batch', batch_id)
nlp = spacy.en.English()
nlp.matcher = None
with open(out_loc, 'wb') as file_:
texts = (strip_meta(text) for text in input_)
texts = (text for text in texts if text.strip())
for doc in nlp.pipe(texts, batch_size=batch_size, n_threads=n_threads):
file_.write(doc.to_bytes())
@plac.annotations(
in_loc=("Location of input file"),
out_dir=("Location of input file"),
n_process=("Number of processes", "option", "p", int),
n_thread=("Number of threads per process", "option", "t", int),
batch_size=("Number of texts to accumulate in a buffer", "option", "b", int)
)
def main(in_loc, out_dir, n_process=1, n_thread=4, batch_size=100):
if not path.exists(out_dir):
path.join(out_dir)
if n_process >= 2:
texts = partition(200000, iter_comments(in_loc))
parallelize(save_parses, enumerate(texts), n_process, [out_dir, n_thread, batch_size],
backend='multiprocessing')
else:
save_parses(0, iter_comments(in_loc), out_dir, n_thread, batch_size)
if __name__ == '__main__':
plac.call(main)

65
examples/pipeline/custom_attr_methods.py
View File

@ -1,35 +1,60 @@
#!/usr/bin/env python
# coding: utf-8
"""This example contains several snippets of methods that can be set via custom
Doc, Token or Span attributes in spaCy v2.0. Attribute methods act like
they're "bound" to the object and are partially applied – i.e. the object
they're called on is passed in as the first argument."""
from __future__ import unicode_literals
they're called on is passed in as the first argument.
* Custom pipeline components: https://alpha.spacy.io//usage/processing-pipelines#custom-components
Developed for: spaCy 2.0.0a17
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
from spacy.lang.en import English
from spacy.tokens import Doc, Span
from spacy import displacy
from pathlib import Path
@plac.annotations(
output_dir=("Output directory for saved HTML", "positional", None, Path))
def main(output_dir=None):
nlp = English() # start off with blank English class
Doc.set_extension('overlap', method=overlap_tokens)
doc1 = nlp(u"Peach emoji is where it has always been.")
doc2 = nlp(u"Peach is the superior emoji.")
print("Text 1:", doc1.text)
print("Text 2:", doc2.text)
print("Overlapping tokens:", doc1._.overlap(doc2))
Doc.set_extension('to_html', method=to_html)
doc = nlp(u"This is a sentence about Apple.")
# add entity manually for demo purposes, to make it work without a model
doc.ents = [Span(doc, 5, 6, label=nlp.vocab.strings['ORG'])]
print("Text:", doc.text)
doc._.to_html(output=output_dir, style='ent')
def to_html(doc, output='/tmp', style='dep'):
"""Doc method extension for saving the current state as a displaCy
visualization.
"""
# generate filename from first six non-punct tokens
file_name = '-'.join([w.text for w in doc[:6] if not w.is_punct]) + '.html'
output_path = Path(output) / file_name
html = displacy.render(doc, style=style, page=True) # render markup
output_path.open('w', encoding='utf-8').write(html) # save to file
print('Saved HTML to {}'.format(output_path))
Doc.set_extension('to_html', method=to_html)
nlp = English()
doc = nlp(u"This is a sentence about Apple.")
# add entity manually for demo purposes, to make it work without a model
doc.ents = [Span(doc, 5, 6, label=nlp.vocab.strings['ORG'])]
doc._.to_html(style='ent')
if output is not None:
output_path = Path(output)
if not output_path.exists():
output_path.mkdir()
output_file = Path(output) / file_name
output_file.open('w', encoding='utf-8').write(html) # save to file
print('Saved HTML to {}'.format(output_file))
else:
print(html)
def overlap_tokens(doc, other_doc):
@ -43,10 +68,10 @@ def overlap_tokens(doc, other_doc):
return overlap
Doc.set_extension('overlap', method=overlap_tokens)
if __name__ == '__main__':
plac.call(main)
nlp = English()
doc1 = nlp(u"Peach emoji is where it has always been.")
doc2 = nlp(u"Peach is the superior emoji.")
tokens = doc1._.overlap(doc2)
print(tokens)
# Expected output:
# Text 1: Peach emoji is where it has always been.
# Text 2: Peach is the superior emoji.
# Overlapping tokens: [Peach, emoji, is, .]

67
examples/pipeline/custom_component_countries_api.py
View File

@ -1,21 +1,45 @@
# coding: utf-8
from __future__ import unicode_literals
#!/usr/bin/env python
# coding: utf8
"""Example of a spaCy v2.0 pipeline component that requests all countries via
the REST Countries API, merges country names into one token, assigns entity
labels and sets attributes on country tokens, e.g. the capital and lat/lng
coordinates. Can be extended with more details from the API.
* REST Countries API: https://restcountries.eu (Mozilla Public License MPL 2.0)
* Custom pipeline components: https://alpha.spacy.io//usage/processing-pipelines#custom-components
Developed for: spaCy 2.0.0a17
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import requests
import plac
from spacy.lang.en import English
from spacy.matcher import PhraseMatcher
from spacy.tokens import Doc, Span, Token
class RESTCountriesComponent(object):
"""Example of a spaCy v2.0 pipeline component that requests all countries
via the REST Countries API, merges country names into one token, assigns
entity labels and sets attributes on country tokens, e.g. the capital and
lat/lng coordinates. Can be extended with more details from the API.
def main():
# For simplicity, we start off with only the blank English Language class
# and no model or pre-defined pipeline loaded.
nlp = English()
rest_countries = RESTCountriesComponent(nlp) # initialise component
nlp.add_pipe(rest_countries) # add it to the pipeline
doc = nlp(u"Some text about Colombia and the Czech Republic")
print('Pipeline', nlp.pipe_names) # pipeline contains component name
print('Doc has countries', doc._.has_country) # Doc contains countries
for token in doc:
if token._.is_country:
print(token.text, token._.country_capital, token._.country_latlng,
token._.country_flag) # country data
print('Entities', [(e.text, e.label_) for e in doc.ents]) # entities
REST Countries API: https://restcountries.eu
API License: Mozilla Public License MPL 2.0
class RESTCountriesComponent(object):
"""spaCy v2.0 pipeline component that requests all countries via
the REST Countries API, merges country names into one token, assigns entity
labels and sets attributes on country tokens.
"""
name = 'rest_countries' # component name, will show up in the pipeline
@ -90,19 +114,12 @@ class RESTCountriesComponent(object):
return any([t._.get('is_country') for t in tokens])
# For simplicity, we start off with only the blank English Language class and
# no model or pre-defined pipeline loaded.
if __name__ == '__main__':
plac.call(main)
nlp = English()
rest_countries = RESTCountriesComponent(nlp) # initialise component
nlp.add_pipe(rest_countries) # add it to the pipeline
doc = nlp(u"Some text about Colombia and the Czech Republic")
print('Pipeline', nlp.pipe_names) # pipeline contains component name
print('Doc has countries', doc._.has_country) # Doc contains countries
for token in doc:
if token._.is_country:
print(token.text, token._.country_capital, token._.country_latlng,
token._.country_flag) # country data
print('Entities', [(e.text, e.label_) for e in doc.ents]) # all countries are entities
# Expected output:
# Pipeline ['rest_countries']
# Doc has countries True
# Colombia Bogotá [4.0, -72.0] https://restcountries.eu/data/col.svg
# Czech Republic Prague [49.75, 15.5] https://restcountries.eu/data/cze.svg
# Entities [('Colombia', 'GPE'), ('Czech Republic', 'GPE')]

62
examples/pipeline/custom_component_entities.py
View File

@ -1,11 +1,45 @@
# coding: utf-8
from __future__ import unicode_literals
#!/usr/bin/env python
# coding: utf8
"""Example of a spaCy v2.0 pipeline component that sets entity annotations
based on list of single or multiple-word company names. Companies are
labelled as ORG and their spans are merged into one token. Additionally,
._.has_tech_org and ._.is_tech_org is set on the Doc/Span and Token
respectively.
* Custom pipeline components: https://alpha.spacy.io//usage/processing-pipelines#custom-components
Developed for: spaCy 2.0.0a17
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
from spacy.lang.en import English
from spacy.matcher import PhraseMatcher
from spacy.tokens import Doc, Span, Token
@plac.annotations(
text=("Text to process", "positional", None, str),
companies=("Names of technology companies", "positional", None, str))
def main(text="Alphabet Inc. is the company behind Google.", *companies):
# For simplicity, we start off with only the blank English Language class
# and no model or pre-defined pipeline loaded.
nlp = English()
if not companies: # set default companies if none are set via args
companies = ['Alphabet Inc.', 'Google', 'Netflix', 'Apple'] # etc.
component = TechCompanyRecognizer(nlp, companies) # initialise component
nlp.add_pipe(component, last=True) # add last to the pipeline
doc = nlp(text)
print('Pipeline', nlp.pipe_names) # pipeline contains component name
print('Tokens', [t.text for t in doc]) # company names from the list are merged
print('Doc has_tech_org', doc._.has_tech_org) # Doc contains tech orgs
print('Token 0 is_tech_org', doc[0]._.is_tech_org) # "Alphabet Inc." is a tech org
print('Token 1 is_tech_org', doc[1]._.is_tech_org) # "is" is not
print('Entities', [(e.text, e.label_) for e in doc.ents]) # all orgs are entities
class TechCompanyRecognizer(object):
"""Example of a spaCy v2.0 pipeline component that sets entity annotations
based on list of single or multiple-word company names. Companies are
@ -67,19 +101,13 @@ class TechCompanyRecognizer(object):
return any([t._.get('is_tech_org') for t in tokens])
# For simplicity, we start off with only the blank English Language class and
# no model or pre-defined pipeline loaded.
if __name__ == '__main__':
plac.call(main)
nlp = English()
companies = ['Alphabet Inc.', 'Google', 'Netflix', 'Apple'] # etc.
component = TechCompanyRecognizer(nlp, companies) # initialise component
nlp.add_pipe(component, last=True) # add it to the pipeline as the last element
doc = nlp(u"Alphabet Inc. is the company behind Google.")
print('Pipeline', nlp.pipe_names) # pipeline contains component name
print('Tokens', [t.text for t in doc]) # company names from the list are merged
print('Doc has_tech_org', doc._.has_tech_org) # Doc contains tech orgs
print('Token 0 is_tech_org', doc[0]._.is_tech_org) # "Alphabet Inc." is a tech org
print('Token 1 is_tech_org', doc[1]._.is_tech_org) # "is" is not
print('Entities', [(e.text, e.label_) for e in doc.ents]) # all orgs are entities
# Expected output:
# Pipeline ['tech_companies']
# Tokens ['Alphabet Inc.', 'is', 'the', 'company', 'behind', 'Google', '.']
# Doc has_tech_org True
# Token 0 is_tech_org True
# Token 1 is_tech_org False
# Entities [('Alphabet Inc.', 'ORG'), ('Google', 'ORG')]

73
examples/pipeline/multi_processing.py Normal file
View File

@ -0,0 +1,73 @@
"""
Example of multi-processing with Joblib. Here, we're exporting
part-of-speech-tagged, true-cased, (very roughly) sentence-separated text, with
each "sentence" on a newline, and spaces between tokens. Data is loaded from
the IMDB movie reviews dataset and will be loaded automatically via Thinc's
built-in dataset loader.
Last updated for: spaCy 2.0.0a18
"""
from __future__ import print_function, unicode_literals
from toolz import partition_all
from pathlib import Path
from joblib import Parallel, delayed
import thinc.extra.datasets
import plac
import spacy
@plac.annotations(
output_dir=("Output directory", "positional", None, Path),
model=("Model name (needs tagger)", "positional", None, str),
n_jobs=("Number of workers", "option", "n", int),
batch_size=("Batch-size for each process", "option", "b", int),
limit=("Limit of entries from the dataset", "option", "l", int))
def main(output_dir, model='en_core_web_sm', n_jobs=4, batch_size=1000,
limit=10000):
nlp = spacy.load(model) # load spaCy model
print("Loaded model '%s'" % model)
if not output_dir.exists():
output_dir.mkdir()
# load and pre-process the IMBD dataset
print("Loading IMDB data...")
data, _ = thinc.extra.datasets.imdb()
texts, _ = zip(*data[-limit:])
partitions = partition_all(batch_size, texts)
items = ((i, [nlp(text) for text in texts], output_dir) for i, texts
in enumerate(partitions))
Parallel(n_jobs=n_jobs)(delayed(transform_texts)(*item) for item in items)
def transform_texts(batch_id, docs, output_dir):
out_path = Path(output_dir) / ('%d.txt' % batch_id)
if out_path.exists(): # return None in case same batch is called again
return None
print('Processing batch', batch_id)
with out_path.open('w', encoding='utf8') as f:
for doc in docs:
f.write(' '.join(represent_word(w) for w in doc if not w.is_space))
f.write('\n')
print('Saved {} texts to {}.txt'.format(len(docs), batch_id))
def represent_word(word):
text = word.text
# True-case, i.e. try to normalize sentence-initial capitals.
# Only do this if the lower-cased form is more probable.
if text.istitle() and is_sent_begin(word) \
and word.prob < word.doc.vocab[text.lower()].prob:
text = text.lower()
return text + '|' + word.tag_
def is_sent_begin(word):
if word.i == 0:
return True
elif word.i >= 2 and word.nbor(-1).text in ('.', '!', '?', '...'):
return True
else:
return False
if __name__ == '__main__':
plac.call(main)

90
examples/pos_tag.py
View File

@ -1,90 +0,0 @@
"""
Print part-of-speech tagged, true-cased, (very roughly) sentence-separated
text, with each "sentence" on a newline, and spaces between tokens. Supports
multi-processing.
"""
from __future__ import print_function, unicode_literals, division
import io
import bz2
import logging
from toolz import partition
from os import path
import spacy.en
from joblib import Parallel, delayed
import plac
import ujson
def parallelize(func, iterator, n_jobs, extra):
extra = tuple(extra)
return Parallel(n_jobs=n_jobs)(delayed(func)(*(item + extra)) for item in iterator)
def iter_texts_from_json_bz2(loc):
"""
Iterator of unicode strings, one per document (here, a comment).
Expects a a path to a BZ2 file, which should be new-line delimited JSON. The
document text should be in a string field titled 'body'.
This is the data format of the Reddit comments corpus.
"""
with bz2.BZ2File(loc) as file_:
for i, line in enumerate(file_):
yield ujson.loads(line)['body']
def transform_texts(batch_id, input_, out_dir):
out_loc = path.join(out_dir, '%d.txt' % batch_id)
if path.exists(out_loc):
return None
print('Batch', batch_id)
nlp = spacy.en.English(parser=False, entity=False)
with io.open(out_loc, 'w', encoding='utf8') as file_:
for text in input_:
doc = nlp(text)
file_.write(' '.join(represent_word(w) for w in doc if not w.is_space))
file_.write('\n')
def represent_word(word):
text = word.text
# True-case, i.e. try to normalize sentence-initial capitals.
# Only do this if the lower-cased form is more probable.
if text.istitle() \
and is_sent_begin(word) \
and word.prob < word.doc.vocab[text.lower()].prob:
text = text.lower()
return text + '|' + word.tag_
def is_sent_begin(word):
# It'd be nice to have some heuristics like these in the library, for these
# times where we don't care so much about accuracy of SBD, and we don't want
# to parse
if word.i == 0:
return True
elif word.i >= 2 and word.nbor(-1).text in ('.', '!', '?', '...'):
return True
else:
return False
@plac.annotations(
in_loc=("Location of input file"),
out_dir=("Location of input file"),
n_workers=("Number of workers", "option", "n", int),
batch_size=("Batch-size for each process", "option", "b", int)
)
def main(in_loc, out_dir, n_workers=4, batch_size=100000):
if not path.exists(out_dir):
path.join(out_dir)
texts = partition(batch_size, iter_texts_from_json_bz2(in_loc))
parallelize(transform_texts, enumerate(texts), n_workers, [out_dir])
if __name__ == '__main__':
plac.call(main)

22
examples/training/load_ner.py
View File

@ -1,22 +0,0 @@
# Load NER
from __future__ import unicode_literals
import spacy
import pathlib
from spacy.pipeline import EntityRecognizer
from spacy.vocab import Vocab
def load_model(model_dir):
model_dir = pathlib.Path(model_dir)
nlp = spacy.load('en', parser=False, entity=False, add_vectors=False)
with (model_dir / 'vocab' / 'strings.json').open('r', encoding='utf8') as file_:
nlp.vocab.strings.load(file_)
nlp.vocab.load_lexemes(model_dir / 'vocab' / 'lexemes.bin')
ner = EntityRecognizer.load(model_dir, nlp.vocab, require=True)
return (nlp, ner)
(nlp, ner) = load_model('ner')
doc = nlp.make_doc('Who is Shaka Khan?')
nlp.tagger(doc)
ner(doc)
for word in doc:
print(word.text, word.orth, word.lower, word.tag_, word.ent_type_, word.ent_iob)

157
examples/training/train_intent_parser.py Normal file
View File

@ -0,0 +1,157 @@
#!/usr/bin/env python
# coding: utf-8
"""Using the parser to recognise your own semantics
spaCy's parser component can be used to trained to predict any type of tree
structure over your input text. You can also predict trees over whole documents
or chat logs, with connections between the sentence-roots used to annotate
discourse structure. In this example, we'll build a message parser for a common
"chat intent": finding local businesses. Our message semantics will have the
following types of relations: ROOT, PLACE, QUALITY, ATTRIBUTE, TIME, LOCATION.
"show me the best hotel in berlin"
('show', 'ROOT', 'show')
('best', 'QUALITY', 'hotel') --> hotel with QUALITY best
('hotel', 'PLACE', 'show') --> show PLACE hotel
('berlin', 'LOCATION', 'hotel') --> hotel with LOCATION berlin
"""
from __future__ import unicode_literals, print_function
import plac
import random
import spacy
from spacy.gold import GoldParse
from spacy.tokens import Doc
from pathlib import Path
# training data: words, head and dependency labels
# for no relation, we simply chose an arbitrary dependency label, e.g. '-'
TRAIN_DATA = [
(
['find', 'a', 'cafe', 'with', 'great', 'wifi'],
[0, 2, 0, 5, 5, 2], # index of token head
['ROOT', '-', 'PLACE', '-', 'QUALITY', 'ATTRIBUTE']
),
(
['find', 'a', 'hotel', 'near', 'the', 'beach'],
[0, 2, 0, 5, 5, 2],
['ROOT', '-', 'PLACE', 'QUALITY', '-', 'ATTRIBUTE']
),
(
['find', 'me', 'the', 'closest', 'gym', 'that', "'s", 'open', 'late'],
[0, 0, 4, 4, 0, 6, 4, 6, 6],
['ROOT', '-', '-', 'QUALITY', 'PLACE', '-', '-', 'ATTRIBUTE', 'TIME']
),
(
['show', 'me', 'the', 'cheapest', 'store', 'that', 'sells', 'flowers'],
[0, 0, 4, 4, 0, 4, 4, 4], # attach "flowers" to store!
['ROOT', '-', '-', 'QUALITY', 'PLACE', '-', '-', 'PRODUCT']
),
(
['find', 'a', 'nice', 'restaurant', 'in', 'london'],
[0, 3, 3, 0, 3, 3],
['ROOT', '-', 'QUALITY', 'PLACE', '-', 'LOCATION']
),
(
['show', 'me', 'the', 'coolest', 'hostel', 'in', 'berlin'],
[0, 0, 4, 4, 0, 4, 4],
['ROOT', '-', '-', 'QUALITY', 'PLACE', '-', 'LOCATION']
),
(
['find', 'a', 'good', 'italian', 'restaurant', 'near', 'work'],
[0, 4, 4, 4, 0, 4, 5],
['ROOT', '-', 'QUALITY', 'ATTRIBUTE', 'PLACE', 'ATTRIBUTE', 'LOCATION']
)
]
@plac.annotations(
model=("Model name. Defaults to blank 'en' model.", "option", "m", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(model=None, output_dir=None, n_iter=100):
"""Load the model, set up the pipeline and train the parser."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
# add the parser to the pipeline if it doesn't exist
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'parser' not in nlp.pipe_names:
parser = nlp.create_pipe('parser')
nlp.add_pipe(parser, first=True)
# otherwise, get it, so we can add labels to it
else:
parser = nlp.get_pipe('parser')
for _, _, deps in TRAIN_DATA:
for dep in deps:
parser.add_label(dep)
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'parser']
with nlp.disable_pipes(*other_pipes): # only train parser
optimizer = nlp.begin_training(lambda: [])
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
for words, heads, deps in TRAIN_DATA:
doc = Doc(nlp.vocab, words=words)
gold = GoldParse(doc, heads=heads, deps=deps)
nlp.update([doc], [gold], sgd=optimizer, losses=losses)
print(losses)
# test the trained model
test_model(nlp)
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
test_model(nlp2)
def test_model(nlp):
texts = ["find a hotel with good wifi",
"find me the cheapest gym near work",
"show me the best hotel in berlin"]
docs = nlp.pipe(texts)
for doc in docs:
print(doc.text)
print([(t.text, t.dep_, t.head.text) for t in doc if t.dep_ != '-'])
if __name__ == '__main__':
plac.call(main)
# Expected output:
# find a hotel with good wifi
# [
# ('find', 'ROOT', 'find'),
# ('hotel', 'PLACE', 'find'),
# ('good', 'QUALITY', 'wifi'),
# ('wifi', 'ATTRIBUTE', 'hotel')
# ]
# find me the cheapest gym near work
# [
# ('find', 'ROOT', 'find'),
# ('cheapest', 'QUALITY', 'gym'),
# ('gym', 'PLACE', 'find')
# ]
# show me the best hotel in berlin
# [
# ('show', 'ROOT', 'show'),
# ('best', 'QUALITY', 'hotel'),
# ('hotel', 'PLACE', 'show'),
# ('berlin', 'LOCATION', 'hotel')
# ]

138
examples/training/train_ner.py
View File

@ -1,13 +1,103 @@
#!/usr/bin/env python
# coding: utf8
"""
Example of training spaCy's named entity recognizer, starting off with an
existing model or a blank model.
For more details, see the documentation:
* Training: https://alpha.spacy.io/usage/training
* NER: https://alpha.spacy.io/usage/linguistic-features#named-entities
Developed for: spaCy 2.0.0a18
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
import random
from pathlib import Path
from spacy.lang.en import English
import spacy
from spacy.gold import GoldParse, biluo_tags_from_offsets
# training data
TRAIN_DATA = [
('Who is Shaka Khan?', [(7, 17, 'PERSON')]),
('I like London and Berlin.', [(7, 13, 'LOC'), (18, 24, 'LOC')])
]
@plac.annotations(
model=("Model name. Defaults to blank 'en' model.", "option", "m", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(model=None, output_dir=None, n_iter=100):
"""Load the model, set up the pipeline and train the entity recognizer."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
# create the built-in pipeline components and add them to the pipeline
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'ner' not in nlp.pipe_names:
ner = nlp.create_pipe('ner')
nlp.add_pipe(ner, last=True)
# function that allows begin_training to get the training data
get_data = lambda: reformat_train_data(nlp.tokenizer, TRAIN_DATA)
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'ner']
with nlp.disable_pipes(*other_pipes): # only train NER
optimizer = nlp.begin_training(get_data)
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
for raw_text, entity_offsets in TRAIN_DATA:
doc = nlp.make_doc(raw_text)
gold = GoldParse(doc, entities=entity_offsets)
nlp.update(
[doc], # Batch of Doc objects
[gold], # Batch of GoldParse objects
drop=0.5, # Dropout -- make it harder to memorise data
sgd=optimizer, # Callable to update weights
losses=losses)
print(losses)
# test the trained model
for text, _ in TRAIN_DATA:
doc = nlp(text)
print('Entities', [(ent.text, ent.label_) for ent in doc.ents])
print('Tokens', [(t.text, t.ent_type_, t.ent_iob) for t in doc])
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
for text, _ in TRAIN_DATA:
doc = nlp2(text)
print('Entities', [(ent.text, ent.label_) for ent in doc.ents])
print('Tokens', [(t.text, t.ent_type_, t.ent_iob) for t in doc])
def reformat_train_data(tokenizer, examples):
"""Reformat data to match JSON format"""
"""Reformat data to match JSON format.
https://alpha.spacy.io/api/annotation#json-input
tokenizer (Tokenizer): Tokenizer to process the raw text.
examples (list): The trainig data.
RETURNS (list): The reformatted training data."""
output = []
for i, (text, entity_offsets) in enumerate(examples):
doc = tokenizer(text)
@ -21,49 +111,5 @@ def reformat_train_data(tokenizer, examples):
return output
def main(model_dir=None):
train_data = [
(
'Who is Shaka Khan?',
[(len('Who is '), len('Who is Shaka Khan'), 'PERSON')]
),
(
'I like London and Berlin.',
[(len('I like '), len('I like London'), 'LOC'),
(len('I like London and '), len('I like London and Berlin'), 'LOC')]
)
]
nlp = English(pipeline=['tensorizer', 'ner'])
get_data = lambda: reformat_train_data(nlp.tokenizer, train_data)
optimizer = nlp.begin_training(get_data)
for itn in range(100):
random.shuffle(train_data)
losses = {}
for raw_text, entity_offsets in train_data:
doc = nlp.make_doc(raw_text)
gold = GoldParse(doc, entities=entity_offsets)
nlp.update(
[doc], # Batch of Doc objects
[gold], # Batch of GoldParse objects
drop=0.5, # Dropout -- make it harder to memorise data
sgd=optimizer, # Callable to update weights
losses=losses)
print(losses)
print("Save to", model_dir)
nlp.to_disk(model_dir)
print("Load from", model_dir)
nlp = spacy.lang.en.English(pipeline=['tensorizer', 'ner'])
nlp.from_disk(model_dir)
for raw_text, _ in train_data:
doc = nlp(raw_text)
for word in doc:
print(word.text, word.ent_type_, word.ent_iob_)
if __name__ == '__main__':
import plac
plac.call(main)
# Who "" 2
# is "" 2
# Shaka "" PERSON 3
# Khan "" PERSON 1
# ? "" 2

206
examples/training/train_ner_standalone.py
View File

@ -1,206 +0,0 @@
#!/usr/bin/env python
'''Example of training a named entity recognition system from scratch using spaCy
This example is written to be self-contained and reasonably transparent.
To achieve that, it duplicates some of spaCy's internal functionality.
Specifically, in this example, we don't use spaCy's built-in Language class to
wire together the Vocab, Tokenizer and EntityRecognizer. Instead, we write
our own simple Pipeline class, so that it's easier to see how the pieces
interact.
Input data:
https://www.lt.informatik.tu-darmstadt.de/fileadmin/user_upload/Group_LangTech/data/GermEval2014_complete_data.zip
Developed for: spaCy 1.7.1
Last tested for: spaCy 2.0.0a13
'''
from __future__ import unicode_literals, print_function
import plac
from pathlib import Path
import random
import json
import tqdm
from thinc.neural.optimizers import Adam
from thinc.neural.ops import NumpyOps
from spacy.vocab import Vocab
from spacy.pipeline import TokenVectorEncoder, NeuralEntityRecognizer
from spacy.tokenizer import Tokenizer
from spacy.tokens import Doc
from spacy.attrs import *
from spacy.gold import GoldParse
from spacy.gold import iob_to_biluo
from spacy.gold import minibatch
from spacy.scorer import Scorer
import spacy.util
try:
unicode
except NameError:
unicode = str
spacy.util.set_env_log(True)
def init_vocab():
return Vocab(
lex_attr_getters={
LOWER: lambda string: string.lower(),
NORM: lambda string: string.lower(),
PREFIX: lambda string: string[0],
SUFFIX: lambda string: string[-3:],
})
class Pipeline(object):
def __init__(self, vocab=None, tokenizer=None, entity=None):
if vocab is None:
vocab = init_vocab()
if tokenizer is None:
tokenizer = Tokenizer(vocab, {}, None, None, None)
if entity is None:
entity = NeuralEntityRecognizer(vocab)
self.vocab = vocab
self.tokenizer = tokenizer
self.entity = entity
self.pipeline = [self.entity]
def begin_training(self):
for model in self.pipeline:
model.begin_training([])
optimizer = Adam(NumpyOps(), 0.001)
return optimizer
def __call__(self, input_):
doc = self.make_doc(input_)
for process in self.pipeline:
process(doc)
return doc
def make_doc(self, input_):
if isinstance(input_, bytes):
input_ = input_.decode('utf8')
if isinstance(input_, unicode):
return self.tokenizer(input_)
else:
return Doc(self.vocab, words=input_)
def make_gold(self, input_, annotations):
doc = self.make_doc(input_)
gold = GoldParse(doc, entities=annotations)
return gold
def update(self, inputs, annots, sgd, losses=None, drop=0.):
if losses is None:
losses = {}
docs = [self.make_doc(input_) for input_ in inputs]
golds = [self.make_gold(input_, annot) for input_, annot in
zip(inputs, annots)]
self.entity.update(docs, golds, drop=drop,
sgd=sgd, losses=losses)
return losses
def evaluate(self, examples):
scorer = Scorer()
for input_, annot in examples:
gold = self.make_gold(input_, annot)
doc = self(input_)
scorer.score(doc, gold)
return scorer.scores
def to_disk(self, path):
path = Path(path)
if not path.exists():
path.mkdir()
elif not path.is_dir():
raise IOError("Can't save pipeline to %s\nNot a directory" % path)
self.vocab.to_disk(path / 'vocab')
self.entity.to_disk(path / 'ner')
def from_disk(self, path):
path = Path(path)
if not path.exists():
raise IOError("Cannot load pipeline from %s\nDoes not exist" % path)
if not path.is_dir():
raise IOError("Cannot load pipeline from %s\nNot a directory" % path)
self.vocab = self.vocab.from_disk(path / 'vocab')
self.entity = self.entity.from_disk(path / 'ner')
def train(nlp, train_examples, dev_examples, nr_epoch=5):
sgd = nlp.begin_training()
print("Iter", "Loss", "P", "R", "F")
for i in range(nr_epoch):
random.shuffle(train_examples)
losses = {}
for batch in minibatch(tqdm.tqdm(train_examples, leave=False), size=8):
inputs, annots = zip(*batch)
nlp.update(list(inputs), list(annots), sgd, losses=losses)
scores = nlp.evaluate(dev_examples)
report_scores(i+1, losses['ner'], scores)
def report_scores(i, loss, scores):
precision = '%.2f' % scores['ents_p']
recall = '%.2f' % scores['ents_r']
f_measure = '%.2f' % scores['ents_f']
print('Epoch %d: %d %s %s %s' % (
i, int(loss), precision, recall, f_measure))
def read_examples(path):
path = Path(path)
with path.open() as file_:
sents = file_.read().strip().split('\n\n')
for sent in sents:
sent = sent.strip()
if not sent:
continue
tokens = sent.split('\n')
while tokens and tokens[0].startswith('#'):
tokens.pop(0)
words = []
iob = []
for token in tokens:
if token.strip():
pieces = token.split('\t')
words.append(pieces[1])
iob.append(pieces[2])
yield words, iob_to_biluo(iob)
def get_labels(examples):
labels = set()
for words, tags in examples:
for tag in tags:
if '-' in tag:
labels.add(tag.split('-')[1])
return sorted(labels)
@plac.annotations(
model_dir=("Path to save the model", "positional", None, Path),
train_loc=("Path to your training data", "positional", None, Path),
dev_loc=("Path to your development data", "positional", None, Path),
)
def main(model_dir, train_loc, dev_loc, nr_epoch=30):
print(model_dir, train_loc, dev_loc)
train_examples = list(read_examples(train_loc))
dev_examples = read_examples(dev_loc)
nlp = Pipeline()
for label in get_labels(train_examples):
nlp.entity.add_label(label)
print("Add label", label)
train(nlp, train_examples, list(dev_examples), nr_epoch)
nlp.to_disk(model_dir)
if __name__ == '__main__':
plac.call(main)

176
examples/training/train_new_entity_type.py
View File

@ -21,104 +21,120 @@ After training your model, you can save it to a directory. We recommend
wrapping models as Python packages, for ease of deployment.
For more details, see the documentation:
* Training the Named Entity Recognizer: https://spacy.io/docs/usage/train-ner
* Saving and loading models: https://spacy.io/docs/usage/saving-loading
* Training: https://alpha.spacy.io/usage/training
* NER: https://alpha.spacy.io/usage/linguistic-features#named-entities
Developed for: spaCy 1.7.6
Last updated for: spaCy 2.0.0a13
Developed for: spaCy 2.0.0a18
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
import random
from pathlib import Path
import random
import spacy
from spacy.gold import GoldParse, minibatch
from spacy.pipeline import NeuralEntityRecognizer
from spacy.pipeline import TokenVectorEncoder
# new entity label
LABEL = 'ANIMAL'
# training data
TRAIN_DATA = [
("Horses are too tall and they pretend to care about your feelings",
[(0, 6, 'ANIMAL')]),
("Do they bite?", []),
("horses are too tall and they pretend to care about your feelings",
[(0, 6, 'ANIMAL')]),
("horses pretend to care about your feelings", [(0, 6, 'ANIMAL')]),
("they pretend to care about your feelings, those horses",
[(48, 54, 'ANIMAL')]),
("horses?", [(0, 6, 'ANIMAL')])
]
@plac.annotations(
model=("Model name. Defaults to blank 'en' model.", "option", "m", str),
new_model_name=("New model name for model meta.", "option", "nm", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(model=None, new_model_name='animal', output_dir=None, n_iter=50):
"""Set up the pipeline and entity recognizer, and train the new entity."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
# Add entity recognizer to model if it's not in the pipeline
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'ner' not in nlp.pipe_names:
ner = nlp.create_pipe('ner')
nlp.add_pipe(ner)
# otherwise, get it, so we can add labels to it
else:
ner = nlp.get_pipe('ner')
ner.add_label(LABEL) # add new entity label to entity recognizer
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'ner']
with nlp.disable_pipes(*other_pipes): # only train NER
random.seed(0)
optimizer = nlp.begin_training(lambda: [])
for itn in range(n_iter):
losses = {}
gold_parses = get_gold_parses(nlp.make_doc, TRAIN_DATA)
for batch in minibatch(gold_parses, size=3):
docs, golds = zip(*batch)
nlp.update(docs, golds, losses=losses, sgd=optimizer,
drop=0.35)
print(losses)
# test the trained model
test_text = 'Do you like horses?'
doc = nlp(test_text)
print("Entities in '%s'" % test_text)
for ent in doc.ents:
print(ent.label_, ent.text)
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.meta['name'] = new_model_name # rename model
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc2 = nlp2(test_text)
for ent in doc2.ents:
print(ent.label_, ent.text)
def get_gold_parses(tokenizer, train_data):
'''Shuffle and create GoldParse objects'''
"""Shuffle and create GoldParse objects.
tokenizer (Tokenizer): Tokenizer to processs the raw text.
train_data (list): The training data.
YIELDS (tuple): (doc, gold) tuples.
"""
random.shuffle(train_data)
for raw_text, entity_offsets in train_data:
doc = tokenizer(raw_text)
gold = GoldParse(doc, entities=entity_offsets)
yield doc, gold
def train_ner(nlp, train_data, output_dir):
random.seed(0)
optimizer = nlp.begin_training(lambda: [])
nlp.meta['name'] = 'en_ent_animal'
for itn in range(50):
losses = {}
for batch in minibatch(get_gold_parses(nlp.make_doc, train_data), size=3):
docs, golds = zip(*batch)
nlp.update(docs, golds, losses=losses, sgd=optimizer, drop=0.35)
print(losses)
if not output_dir:
return
elif not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
def main(model_name, output_directory=None):
print("Creating initial model", model_name)
nlp = spacy.blank(model_name)
if output_directory is not None:
output_directory = Path(output_directory)
train_data = [
(
"Horses are too tall and they pretend to care about your feelings",
[(0, 6, 'ANIMAL')],
),
(
"Do they bite?",
[],
),
(
"horses are too tall and they pretend to care about your feelings",
[(0, 6, 'ANIMAL')]
),
(
"horses pretend to care about your feelings",
[(0, 6, 'ANIMAL')]
),
(
"they pretend to care about your feelings, those horses",
[(48, 54, 'ANIMAL')]
),
(
"horses?",
[(0, 6, 'ANIMAL')]
)
]
nlp.add_pipe(TokenVectorEncoder(nlp.vocab))
ner = NeuralEntityRecognizer(nlp.vocab)
ner.add_label('ANIMAL')
nlp.add_pipe(ner)
train_ner(nlp, train_data, output_directory)
# Test that the entity is recognized
text = 'Do you like horses?'
print("Ents in 'Do you like horses?':")
doc = nlp(text)
for ent in doc.ents:
print(ent.label_, ent.text)
if output_directory:
print("Loading from", output_directory)
nlp2 = spacy.load(output_directory)
doc2 = nlp2('Do you like horses?')
for ent in doc2.ents:
print(ent.label_, ent.text)
if __name__ == '__main__':
import plac
plac.call(main)

150
examples/training/train_parser.py
View File

@ -1,75 +1,109 @@
#!/usr/bin/env python
# coding: utf8
"""
Example of training spaCy dependency parser, starting off with an existing model
or a blank model.
For more details, see the documentation:
* Training: https://alpha.spacy.io/usage/training
* Dependency Parse: https://alpha.spacy.io/usage/linguistic-features#dependency-parse
Developed for: spaCy 2.0.0a18
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import json
import pathlib
import plac
import random
from pathlib import Path
import spacy
from spacy.pipeline import DependencyParser
from spacy.gold import GoldParse
from spacy.tokens import Doc
def train_parser(nlp, train_data, left_labels, right_labels):
parser = DependencyParser(
nlp.vocab,
left_labels=left_labels,
right_labels=right_labels)
for itn in range(1000):
random.shuffle(train_data)
loss = 0
for words, heads, deps in train_data:
doc = Doc(nlp.vocab, words=words)
gold = GoldParse(doc, heads=heads, deps=deps)
loss += parser.update(doc, gold)
parser.model.end_training()
return parser
# training data
TRAIN_DATA = [
(
['They', 'trade', 'mortgage', '-', 'backed', 'securities', '.'],
[1, 1, 4, 4, 5, 1, 1],
['nsubj', 'ROOT', 'compound', 'punct', 'nmod', 'dobj', 'punct']
),
(
['I', 'like', 'London', 'and', 'Berlin', '.'],
[1, 1, 1, 2, 2, 1],
['nsubj', 'ROOT', 'dobj', 'cc', 'conj', 'punct']
)
]
def main(model_dir=None):
if model_dir is not None:
model_dir = pathlib.Path(model_dir)
if not model_dir.exists():
model_dir.mkdir()
assert model_dir.is_dir()
@plac.annotations(
model=("Model name. Defaults to blank 'en' model.", "option", "m", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(model=None, output_dir=None, n_iter=1000):
"""Load the model, set up the pipeline and train the parser."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
nlp = spacy.load('en', tagger=False, parser=False, entity=False, add_vectors=False)
# add the parser to the pipeline if it doesn't exist
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'parser' not in nlp.pipe_names:
parser = nlp.create_pipe('parser')
nlp.add_pipe(parser, first=True)
# otherwise, get it, so we can add labels to it
else:
parser = nlp.get_pipe('parser')
train_data = [
(
['They', 'trade', 'mortgage', '-', 'backed', 'securities', '.'],
[1, 1, 4, 4, 5, 1, 1],
['nsubj', 'ROOT', 'compound', 'punct', 'nmod', 'dobj', 'punct']
),
(
['I', 'like', 'London', 'and', 'Berlin', '.'],
[1, 1, 1, 2, 2, 1],
['nsubj', 'ROOT', 'dobj', 'cc', 'conj', 'punct']
)
]
left_labels = set()
right_labels = set()
for _, heads, deps in train_data:
for i, (head, dep) in enumerate(zip(heads, deps)):
if i < head:
left_labels.add(dep)
elif i > head:
right_labels.add(dep)
parser = train_parser(nlp, train_data, sorted(left_labels), sorted(right_labels))
# add labels to the parser
for _, _, deps in TRAIN_DATA:
for dep in deps:
parser.add_label(dep)
doc = Doc(nlp.vocab, words=['I', 'like', 'securities', '.'])
parser(doc)
for word in doc:
print(word.text, word.dep_, word.head.text)
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'parser']
with nlp.disable_pipes(*other_pipes): # only train parser
optimizer = nlp.begin_training(lambda: [])
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
for words, heads, deps in TRAIN_DATA:
doc = Doc(nlp.vocab, words=words)
gold = GoldParse(doc, heads=heads, deps=deps)
nlp.update([doc], [gold], sgd=optimizer, losses=losses)
print(losses)
if model_dir is not None:
with (model_dir / 'config.json').open('w') as file_:
json.dump(parser.cfg, file_)
parser.model.dump(str(model_dir / 'model'))
# test the trained model
test_text = "I like securities."
doc = nlp(test_text)
print('Dependencies', [(t.text, t.dep_, t.head.text) for t in doc])
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc = nlp2(test_text)
print('Dependencies', [(t.text, t.dep_, t.head.text) for t in doc])
if __name__ == '__main__':
main()
# I nsubj like
# like ROOT like
# securities dobj like
# . cc securities
plac.call(main)
# expected result:
# [
# ('I', 'nsubj', 'like'),
# ('like', 'ROOT', 'like'),
# ('securities', 'dobj', 'like'),
# ('.', 'punct', 'like')
# ]

115
examples/training/train_tagger.py
View File

@ -1,18 +1,28 @@
"""A quick example for training a part-of-speech tagger, without worrying
about the tokenization, or other language-specific customizations."""
#!/usr/bin/env python
# coding: utf8
"""
A simple example for training a part-of-speech tagger with a custom tag map.
To allow us to update the tag map with our custom one, this example starts off
with a blank Language class and modifies its defaults.
from __future__ import unicode_literals
from __future__ import print_function
For more details, see the documentation:
* Training: https://alpha.spacy.io/usage/training
* POS Tagging: https://alpha.spacy.io/usage/linguistic-features#pos-tagging
Developed for: spaCy 2.0.0a18
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
import random
from pathlib import Path
from spacy.vocab import Vocab
from spacy.tagger import Tagger
import spacy
from spacy.util import get_lang_class
from spacy.tokens import Doc
from spacy.gold import GoldParse
import random
# You need to define a mapping from your data's part-of-speech tag names to the
# Universal Part-of-Speech tag set, as spaCy includes an enum of these tags.
@ -28,54 +38,67 @@ TAG_MAP = {
# Usually you'll read this in, of course. Data formats vary.
# Ensure your strings are unicode.
DATA = [
(
["I", "like", "green", "eggs"],
["N", "V", "J", "N"]
),
(
["Eat", "blue", "ham"],
["V", "J", "N"]
)
TRAIN_DATA = [
(["I", "like", "green", "eggs"], ["N", "V", "J", "N"]),
(["Eat", "blue", "ham"], ["V", "J", "N"])
]
def ensure_dir(path):
if not path.exists():
path.mkdir()
@plac.annotations(
lang=("ISO Code of language to use", "option", "l", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(lang='en', output_dir=None, n_iter=25):
"""Create a new model, set up the pipeline and train the tagger. In order to
train the tagger with a custom tag map, we're creating a new Language
instance with a custom vocab.
"""
lang_cls = get_lang_class(lang) # get Language class
lang_cls.Defaults.tag_map.update(TAG_MAP) # add tag map to defaults
nlp = lang_cls() # initialise Language class
# add the tagger to the pipeline
# nlp.create_pipe works for built-ins that are registered with spaCy
tagger = nlp.create_pipe('tagger')
nlp.add_pipe(tagger)
def main(output_dir=None):
optimizer = nlp.begin_training(lambda: [])
for i in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
for words, tags in TRAIN_DATA:
doc = Doc(nlp.vocab, words=words)
gold = GoldParse(doc, tags=tags)
nlp.update([doc], [gold], sgd=optimizer, losses=losses)
print(losses)
# test the trained model
test_text = "I like blue eggs"
doc = nlp(test_text)
print('Tags', [(t.text, t.tag_, t.pos_) for t in doc])
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
ensure_dir(output_dir)
ensure_dir(output_dir / "pos")
ensure_dir(output_dir / "vocab")
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
vocab = Vocab(tag_map=TAG_MAP)
# The default_templates argument is where features are specified. See
# spacy/tagger.pyx for the defaults.
tagger = Tagger(vocab)
for i in range(25):
for words, tags in DATA:
doc = Doc(vocab, words=words)
gold = GoldParse(doc, tags=tags)
tagger.update(doc, gold)
random.shuffle(DATA)
tagger.model.end_training()
doc = Doc(vocab, orths_and_spaces=zip(["I", "like", "blue", "eggs"], [True] * 4))
tagger(doc)
for word in doc:
print(word.text, word.tag_, word.pos_)
if output_dir is not None:
tagger.model.dump(str(output_dir / 'pos' / 'model'))
with (output_dir / 'vocab' / 'strings.json').open('w') as file_:
tagger.vocab.strings.dump(file_)
# test the save model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc = nlp2(test_text)
print('Tags', [(t.text, t.tag_, t.pos_) for t in doc])
if __name__ == '__main__':
plac.call(main)
# I V VERB
# like V VERB
# blue N NOUN
# eggs N NOUN
# Expected output:
# [
# ('I', 'N', 'NOUN'),
# ('like', 'V', 'VERB'),
# ('blue', 'J', 'ADJ'),
# ('eggs', 'N', 'NOUN')
# ]

188
examples/training/train_textcat.py
View File

@ -1,58 +1,119 @@
'''Train a multi-label convolutional neural network text classifier,
using the spacy.pipeline.TextCategorizer component. The model is then added
to spacy.pipeline, and predictions are available at `doc.cats`.
'''
from __future__ import unicode_literals
#!/usr/bin/env python
# coding: utf8
"""Train a multi-label convolutional neural network text classifier on the
IMDB dataset, using the TextCategorizer component. The dataset will be loaded
automatically via Thinc's built-in dataset loader. The model is added to
spacy.pipeline, and predictions are available via `doc.cats`.
For more details, see the documentation:
* Training: https://alpha.spacy.io/usage/training
* Text classification: https://alpha.spacy.io/usage/text-classification
Developed for: spaCy 2.0.0a18
Last updated for: spaCy 2.0.0a18
"""
from __future__ import unicode_literals, print_function
import plac
import random
import tqdm
from thinc.neural.optimizers import Adam
from thinc.neural.ops import NumpyOps
from pathlib import Path
import thinc.extra.datasets
import spacy.lang.en
import spacy
from spacy.gold import GoldParse, minibatch
from spacy.util import compounding
from spacy.pipeline import TextCategorizer
# TODO: Remove this once we're not supporting models trained with thinc <6.9.0
import thinc.neural._classes.layernorm
thinc.neural._classes.layernorm.set_compat_six_eight(False)
@plac.annotations(
model=("Model name. Defaults to blank 'en' model.", "option", "m", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(model=None, output_dir=None, n_iter=20):
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
def train_textcat(tokenizer, textcat,
train_texts, train_cats, dev_texts, dev_cats,
n_iter=20):
'''
Train the TextCategorizer without associated pipeline.
'''
textcat.begin_training()
optimizer = Adam(NumpyOps(), 0.001)
train_docs = [tokenizer(text) for text in train_texts]
# add the text classifier to the pipeline if it doesn't exist
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'textcat' not in nlp.pipe_names:
# textcat = nlp.create_pipe('textcat')
textcat = TextCategorizer(nlp.vocab, labels=['POSITIVE'])
nlp.add_pipe(textcat, last=True)
# otherwise, get it, so we can add labels to it
else:
textcat = nlp.get_pipe('textcat')
# add label to text classifier
# textcat.add_label('POSITIVE')
# load the IMBD dataset
print("Loading IMDB data...")
(train_texts, train_cats), (dev_texts, dev_cats) = load_data(limit=2000)
train_docs = [nlp.tokenizer(text) for text in train_texts]
train_gold = [GoldParse(doc, cats=cats) for doc, cats in
zip(train_docs, train_cats)]
train_data = list(zip(train_docs, train_gold))
batch_sizes = compounding(4., 128., 1.001)
for i in range(n_iter):
losses = {}
# Progress bar and minibatching
batches = minibatch(tqdm.tqdm(train_data, leave=False), size=batch_sizes)
for batch in batches:
docs, golds = zip(*batch)
textcat.update(docs, golds, sgd=optimizer, drop=0.2,
losses=losses)
with textcat.model.use_params(optimizer.averages):
scores = evaluate(tokenizer, textcat, dev_texts, dev_cats)
yield losses['textcat'], scores
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'textcat']
with nlp.disable_pipes(*other_pipes): # only train textcat
optimizer = nlp.begin_training(lambda: [])
print("Training the model...")
print('{:^5}\t{:^5}\t{:^5}\t{:^5}'.format('LOSS', 'P', 'R', 'F'))
for i in range(n_iter):
losses = {}
# batch up the examples using spaCy's minibatch
batches = minibatch(train_data, size=compounding(4., 128., 1.001))
for batch in batches:
docs, golds = zip(*batch)
nlp.update(docs, golds, sgd=optimizer, drop=0.2, losses=losses)
with textcat.model.use_params(optimizer.averages):
# evaluate on the dev data split off in load_data()
scores = evaluate(nlp.tokenizer, textcat, dev_texts, dev_cats)
print('{0:.3f}\t{0:.3f}\t{0:.3f}\t{0:.3f}' # print a simple table
.format(losses['textcat'], scores['textcat_p'],
scores['textcat_r'], scores['textcat_f']))
# test the trained model
test_text = "This movie sucked"
doc = nlp(test_text)
print(test_text, doc.cats)
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc2 = nlp2(test_text)
print(test_text, doc2.cats)
def load_data(limit=0, split=0.8):
"""Load data from the IMDB dataset."""
# Partition off part of the train data for evaluation
train_data, _ = thinc.extra.datasets.imdb()
random.shuffle(train_data)
train_data = train_data[-limit:]
texts, labels = zip(*train_data)
cats = [{'POSITIVE': bool(y)} for y in labels]
split = int(len(train_data) * split)
return (texts[:split], cats[:split]), (texts[split:], cats[split:])
def evaluate(tokenizer, textcat, texts, cats):
docs = (tokenizer(text) for text in texts)
tp = 1e-8 # True positives
fp = 1e-8 # False positives
fn = 1e-8 # False negatives
tn = 1e-8 # True negatives
tp = 1e-8 # True positives
fp = 1e-8 # False positives
fn = 1e-8 # False negatives
tn = 1e-8 # True negatives
for i, doc in enumerate(textcat.pipe(docs)):
gold = cats[i]
for label, score in doc.cats.items():
@ -66,55 +127,10 @@ def evaluate(tokenizer, textcat, texts, cats):
tn += 1
elif score < 0.5 and gold[label] >= 0.5:
fn += 1
precis = tp / (tp + fp)
precision = tp / (tp + fp)
recall = tp / (tp + fn)
fscore = 2 * (precis * recall) / (precis + recall)
return {'textcat_p': precis, 'textcat_r': recall, 'textcat_f': fscore}
def load_data(limit=0):
# Partition off part of the train data --- avoid running experiments
# against test.
train_data, _ = thinc.extra.datasets.imdb()
random.shuffle(train_data)
train_data = train_data[-limit:]
texts, labels = zip(*train_data)
cats = [{'POSITIVE': bool(y)} for y in labels]
split = int(len(train_data) * 0.8)
train_texts = texts[:split]
train_cats = cats[:split]
dev_texts = texts[split:]
dev_cats = cats[split:]
return (train_texts, train_cats), (dev_texts, dev_cats)
def main(model_loc=None):
nlp = spacy.lang.en.English()
tokenizer = nlp.tokenizer
textcat = TextCategorizer(tokenizer.vocab, labels=['POSITIVE'])
print("Load IMDB data")
(train_texts, train_cats), (dev_texts, dev_cats) = load_data(limit=2000)
print("Itn.\tLoss\tP\tR\tF")
progress = '{i:d} {loss:.3f} {textcat_p:.3f} {textcat_r:.3f} {textcat_f:.3f}'
for i, (loss, scores) in enumerate(train_textcat(tokenizer, textcat,
train_texts, train_cats,
dev_texts, dev_cats, n_iter=20)):
print(progress.format(i=i, loss=loss, **scores))
# How to save, load and use
nlp.pipeline.append(textcat)
if model_loc is not None:
nlp.to_disk(model_loc)
nlp = spacy.load(model_loc)
doc = nlp(u'This movie sucked!')
print(doc.cats)
f_score = 2 * (precision * recall) / (precision + recall)
return {'textcat_p': precision, 'textcat_r': recall, 'textcat_f': f_score}
if __name__ == '__main__':

36
examples/twitter_filter.py
View File

@ -1,36 +0,0 @@
# encoding: utf8
from __future__ import unicode_literals, print_function
import plac
import codecs
import pathlib
import random
import twython
import spacy.en
import _handler
class Connection(twython.TwythonStreamer):
def __init__(self, keys_dir, nlp, query):
keys_dir = pathlib.Path(keys_dir)
read = lambda fn: (keys_dir / (fn + '.txt')).open().read().strip()
api_key = map(read, ['key', 'secret', 'token', 'token_secret'])
twython.TwythonStreamer.__init__(self, *api_key)
self.nlp = nlp
self.query = query
def on_success(self, data):
_handler.handle_tweet(self.nlp, data, self.query)
if random.random() >= 0.1:
reload(_handler)
def main(keys_dir, term):
nlp = spacy.en.English()
twitter = Connection(keys_dir, nlp, term)
twitter.statuses.filter(track=term, language='en')
if __name__ == '__main__':
plac.call(main)

15
examples/vectors_fast_text.py
View File

@ -1,16 +1,19 @@
'''Load vectors for a language trained using FastText
#!/usr/bin/env python
# coding: utf8
"""Load vectors for a language trained using fastText
https://github.com/facebookresearch/fastText/blob/master/pretrained-vectors.md
'''
"""
from __future__ import unicode_literals
import plac
import numpy
import spacy.language
import from spacy.language import Language
@plac.annotations(
vectors_loc=("Path to vectors", "positional", None, str))
def main(vectors_loc):
nlp = spacy.language.Language()
nlp = Language()
with open(vectors_loc, 'rb') as file_:
header = file_.readline()
@ -18,7 +21,7 @@ def main(vectors_loc):
nlp.vocab.clear_vectors(int(nr_dim))
for line in file_:
line = line.decode('utf8')
pieces = line.split()
pieces = line.split()
word = pieces[0]
vector = numpy.asarray([float(v) for v in pieces[1:]], dtype='f')
nlp.vocab.set_vector(word, vector)

5
setup.py
View File

@ -30,19 +30,14 @@ MOD_NAMES = [
'spacy.syntax._state',
'spacy.syntax._beam_utils',
'spacy.tokenizer',
'spacy._cfile',
'spacy.syntax.parser',
'spacy.syntax.nn_parser',
'spacy.syntax.beam_parser',
'spacy.syntax.nonproj',
'spacy.syntax.transition_system',
'spacy.syntax.arc_eager',
'spacy.syntax._parse_features',
'spacy.gold',
'spacy.tokens.doc',
'spacy.tokens.span',
'spacy.tokens.token',
'spacy.cfile',
'spacy.matcher',
'spacy.syntax.ner',
'spacy.symbols',

26
spacy/_cfile.pxd
View File

@ -1,26 +0,0 @@
from libc.stdio cimport fopen, fclose, fread, fwrite, FILE
from cymem.cymem cimport Pool
cdef class CFile:
cdef FILE* fp
cdef bint is_open
cdef Pool mem
cdef int size # For compatibility with subclass
cdef int _capacity # For compatibility with subclass
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1
cdef int write_from(self, void* src, size_t number, size_t elem_size) except -1
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *
cdef class StringCFile(CFile):
cdef unsigned char* data
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1
cdef int write_from(self, void* src, size_t number, size_t elem_size) except -1
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *

88
spacy/_cfile.pyx
View File

@ -1,88 +0,0 @@
from libc.stdio cimport fopen, fclose, fread, fwrite, FILE
from libc.string cimport memcpy
cdef class CFile:
def __init__(self, loc, mode, on_open_error=None):
if isinstance(mode, unicode):
mode_str = mode.encode('ascii')
else:
mode_str = mode
if hasattr(loc, 'as_posix'):
loc = loc.as_posix()
self.mem = Pool()
cdef bytes bytes_loc = loc.encode('utf8') if type(loc) == unicode else loc
self.fp = fopen(<char*>bytes_loc, mode_str)
if self.fp == NULL:
if on_open_error is not None:
on_open_error()
else:
raise IOError("Could not open binary file %s" % bytes_loc)
self.is_open = True
def __dealloc__(self):
if self.is_open:
fclose(self.fp)
def close(self):
fclose(self.fp)
self.is_open = False
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1:
st = fread(dest, elem_size, number, self.fp)
if st != number:
raise IOError
cdef int write_from(self, void* src, size_t number, size_t elem_size) except -1:
st = fwrite(src, elem_size, number, self.fp)
if st != number:
raise IOError
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *:
cdef void* dest = mem.alloc(number, elem_size)
self.read_into(dest, number, elem_size)
return dest
def write_unicode(self, unicode value):
cdef bytes py_bytes = value.encode('utf8')
cdef char* chars = <char*>py_bytes
self.write(sizeof(char), len(py_bytes), chars)
cdef class StringCFile:
def __init__(self, mode, bytes data=b'', on_open_error=None):
self.mem = Pool()
self.is_open = 'w' in mode
self._capacity = max(len(data), 8)
self.size = len(data)
self.data = <unsigned char*>self.mem.alloc(1, self._capacity)
for i in range(len(data)):
self.data[i] = data[i]
def close(self):
self.is_open = False
def string_data(self):
return (self.data-self.size)[:self.size]
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1:
memcpy(dest, self.data, elem_size * number)
self.data += elem_size * number
cdef int write_from(self, void* src, size_t elem_size, size_t number) except -1:
write_size = number * elem_size
if (self.size + write_size) >= self._capacity:
self._capacity = (self.size + write_size) * 2
self.data = <unsigned char*>self.mem.realloc(self.data, self._capacity)
memcpy(&self.data[self.size], src, elem_size * number)
self.size += write_size
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *:
cdef void* dest = mem.alloc(number, elem_size)
self.read_into(dest, number, elem_size)
return dest
def write_unicode(self, unicode value):
cdef bytes py_bytes = value.encode('utf8')
cdef char* chars = <char*>py_bytes
self.write(sizeof(char), len(py_bytes), chars)

2
spacy/_ml.py
View File

@ -96,7 +96,6 @@ def _zero_init(model):
@layerize
def _preprocess_doc(docs, drop=0.):
keys = [doc.to_array([LOWER]) for doc in docs]
keys = [a[:, 0] for a in keys]
ops = Model.ops
lengths = ops.asarray([arr.shape[0] for arr in keys])
keys = ops.xp.concatenate(keys)
@ -464,7 +463,6 @@ def zero_init(model):
@layerize
def preprocess_doc(docs, drop=0.):
keys = [doc.to_array([LOWER]) for doc in docs]
keys = [a[:, 0] for a in keys]
ops = Model.ops
lengths = ops.asarray([arr.shape[0] for arr in keys])
keys = ops.xp.concatenate(keys)

33
spacy/cfile.pxd
View File

@ -1,33 +0,0 @@
from libc.stdio cimport fopen, fclose, fread, fwrite, FILE
from cymem.cymem cimport Pool
cdef class CFile:
cdef FILE* fp
cdef unsigned char* data
cdef int is_open
cdef Pool mem
cdef int size # For compatibility with subclass
cdef int i # For compatibility with subclass
cdef int _capacity # For compatibility with subclass
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1
cdef int write_from(self, void* src, size_t number, size_t elem_size) except -1
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *
cdef class StringCFile:
cdef unsigned char* data
cdef int is_open
cdef Pool mem
cdef int size # For compatibility with subclass
cdef int i # For compatibility with subclass
cdef int _capacity # For compatibility with subclass
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1
cdef int write_from(self, void* src, size_t number, size_t elem_size) except -1
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *

103
spacy/cfile.pyx
View File

@ -1,103 +0,0 @@
# coding: utf8
from __future__ import unicode_literals
from libc.stdio cimport fopen, fclose, fread, fwrite
from libc.string cimport memcpy
cdef class CFile:
def __init__(self, loc, mode, on_open_error=None):
if isinstance(mode, unicode):
mode_str = mode.encode('ascii')
else:
mode_str = mode
if hasattr(loc, 'as_posix'):
loc = loc.as_posix()
self.mem = Pool()
cdef bytes bytes_loc = loc.encode('utf8') if type(loc) == unicode else loc
self.fp = fopen(<char*>bytes_loc, mode_str)
if self.fp == NULL:
if on_open_error is not None:
on_open_error()
else:
raise IOError("Could not open binary file %s" % bytes_loc)
self.is_open = True
def __dealloc__(self):
if self.is_open:
fclose(self.fp)
def close(self):
fclose(self.fp)
self.is_open = False
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1:
st = fread(dest, elem_size, number, self.fp)
if st != number:
raise IOError
cdef int write_from(self, void* src, size_t number, size_t elem_size) except -1:
st = fwrite(src, elem_size, number, self.fp)
if st != number:
raise IOError
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *:
cdef void* dest = mem.alloc(number, elem_size)
self.read_into(dest, number, elem_size)
return dest
def write_unicode(self, unicode value):
cdef bytes py_bytes = value.encode('utf8')
cdef char* chars = <char*>py_bytes
self.write(sizeof(char), len(py_bytes), chars)
cdef class StringCFile:
def __init__(self, bytes data, mode, on_open_error=None):
self.mem = Pool()
self.is_open = 1 if 'w' in mode else 0
self._capacity = max(len(data), 8)
self.size = len(data)
self.i = 0
self.data = <unsigned char*>self.mem.alloc(1, self._capacity)
for i in range(len(data)):
self.data[i] = data[i]
def __dealloc__(self):
# Important to override this -- or
# we try to close a non-existant file pointer!
pass
def close(self):
self.is_open = False
def string_data(self):
cdef bytes byte_string = b'\0' * (self.size)
bytes_ptr = <char*>byte_string
for i in range(self.size):
bytes_ptr[i] = self.data[i]
print(byte_string)
return byte_string
cdef int read_into(self, void* dest, size_t number, size_t elem_size) except -1:
if self.i+(number * elem_size) < self.size:
memcpy(dest, &self.data[self.i], elem_size * number)
self.i += elem_size * number
cdef int write_from(self, void* src, size_t elem_size, size_t number) except -1:
write_size = number * elem_size
if (self.size + write_size) >= self._capacity:
self._capacity = (self.size + write_size) * 2
self.data = <unsigned char*>self.mem.realloc(self.data, self._capacity)
memcpy(&self.data[self.size], src, write_size)
self.size += write_size
cdef void* alloc_read(self, Pool mem, size_t number, size_t elem_size) except *:
cdef void* dest = mem.alloc(number, elem_size)
self.read_into(dest, number, elem_size)
return dest
def write_unicode(self, unicode value):
cdef bytes py_bytes = value.encode('utf8')
cdef char* chars = <char*>py_bytes
self.write(sizeof(char), len(py_bytes), chars)

29
spacy/cli/package.py
View File

@ -43,7 +43,7 @@ def package(cmd, input_dir, output_dir, meta_path=None, create_meta=False, force
prints(meta_path, title="Reading meta.json from file")
meta = util.read_json(meta_path)
else:
meta = generate_meta()
meta = generate_meta(input_dir)
meta = validate_meta(meta, ['lang', 'name', 'version'])
model_name = meta['lang'] + '_' + meta['name']
@ -77,7 +77,8 @@ def create_file(file_path, contents):
file_path.open('w', encoding='utf-8').write(contents)
def generate_meta():
def generate_meta(model_path):
meta = {}
settings = [('lang', 'Model language', 'en'),
('name', 'Model name', 'model'),
('version', 'Model version', '0.0.0'),
@ -87,31 +88,21 @@ def generate_meta():
('email', 'Author email', False),
('url', 'Author website', False),
('license', 'License', 'CC BY-NC 3.0')]
prints("Enter the package settings for your model.", title="Generating meta.json")
meta = {}
nlp = util.load_model_from_path(Path(model_path))
meta['pipeline'] = nlp.pipe_names
meta['vectors'] = {'width': nlp.vocab.vectors_length,
'entries': len(nlp.vocab.vectors)}
prints("Enter the package settings for your model. The following "
"information will be read from your model data: pipeline, vectors.",
title="Generating meta.json")
for setting, desc, default in settings:
response = util.get_raw_input(desc, default)
meta[setting] = default if response == '' and default else response
meta['pipeline'] = generate_pipeline()
if about.__title__ != 'spacy':
meta['parent_package'] = about.__title__
return meta
def generate_pipeline():
prints("If set to 'True', the default pipeline is used. If set to 'False', "
"the pipeline will be disabled. Components should be specified as a "
"comma-separated list of component names, e.g. tagger, "
"parser, ner. For more information, see the docs on processing pipelines.",
title="Enter your model's pipeline components")
pipeline = util.get_raw_input("Pipeline components", True)
subs = {'True': True, 'False': False}
if pipeline in subs:
return subs[pipeline]
else:
return [p.strip() for p in pipeline.split(',')]
def validate_meta(meta, keys):
for key in keys:
if key not in meta or meta[key] == '':

5
spacy/cli/train.py
View File

@ -144,7 +144,10 @@ def train(cmd, lang, output_dir, train_data, dev_data, n_iter=30, n_sents=0,
file_.write(json_dumps(scorer.scores))
meta_loc = output_path / ('model%d' % i) / 'meta.json'
meta['accuracy'] = scorer.scores
meta['speed'] = {'nwords': nwords, 'cpu':cpu_wps, 'gpu': gpu_wps}
meta['speed'] = {'nwords': nwords, 'cpu': cpu_wps,
'gpu': gpu_wps}
meta['vectors'] = {'width': nlp.vocab.vectors_length,
'entries': len(nlp.vocab.vectors)}
meta['lang'] = nlp.lang
meta['pipeline'] = pipeline
meta['spacy_version'] = '>=%s' % about.__version__

11
spacy/glossary.py
View File

@ -3,6 +3,16 @@ from __future__ import unicode_literals
def explain(term):
"""Get a description for a given POS tag, dependency label or entity type.
term (unicode): The term to explain.
RETURNS (unicode): The explanation, or `None` if not found in the glossary.
EXAMPLE:
>>> spacy.explain(u'NORP')
>>> doc = nlp(u'Hello world')
>>> print([w.text, w.tag_, spacy.explain(w.tag_) for w in doc])
"""
if term in GLOSSARY:
return GLOSSARY[term]
@ -283,6 +293,7 @@ GLOSSARY = {
'PRODUCT': 'Objects, vehicles, foods, etc. (not services)',
'EVENT': 'Named hurricanes, battles, wars, sports events, etc.',
'WORK_OF_ART': 'Titles of books, songs, etc.',
'LAW': 'Named documents made into laws.',
'LANGUAGE': 'Any named language',
'DATE': 'Absolute or relative dates or periods',
'TIME': 'Times smaller than a day',

76
spacy/language.py
View File

@ -1,6 +1,7 @@
# coding: utf8
from __future__ import absolute_import, unicode_literals
from contextlib import contextmanager
import copy
from thinc.neural import Model
import random
@ -15,10 +16,9 @@ from .tokenizer import Tokenizer
from .vocab import Vocab
from .tagger import Tagger
from .lemmatizer import Lemmatizer
from .syntax.parser import get_templates
from .pipeline import NeuralDependencyParser, TokenVectorEncoder, NeuralTagger
from .pipeline import NeuralEntityRecognizer, SimilarityHook, TextCategorizer
from .pipeline import DependencyParser, Tensorizer, Tagger
from .pipeline import EntityRecognizer, SimilarityHook, TextCategorizer
from .compat import Optimizer
from .compat import json_dumps, izip, copy_reg
@ -75,9 +75,6 @@ class BaseDefaults(object):
infixes = tuple(TOKENIZER_INFIXES)
tag_map = dict(TAG_MAP)
tokenizer_exceptions = {}
parser_features = get_templates('parser')
entity_features = get_templates('ner')
tagger_features = Tagger.feature_templates # TODO -- fix this
stop_words = set()
lemma_rules = {}
lemma_exc = {}
@ -102,9 +99,9 @@ class Language(object):
factories = {
'tokenizer': lambda nlp: nlp.Defaults.create_tokenizer(nlp),
'tensorizer': lambda nlp, **cfg: TokenVectorEncoder(nlp.vocab, **cfg),
'tagger': lambda nlp, **cfg: NeuralTagger(nlp.vocab, **cfg),
'parser': lambda nlp, **cfg: NeuralDependencyParser(nlp.vocab, **cfg),
'ner': lambda nlp, **cfg: NeuralEntityRecognizer(nlp.vocab, **cfg),
'tagger': lambda nlp, **cfg: Tagger(nlp.vocab, **cfg),
'parser': lambda nlp, **cfg: DependencyParser(nlp.vocab, **cfg),
'ner': lambda nlp, **cfg: EntityRecognizer(nlp.vocab, **cfg),
'similarity': lambda nlp, **cfg: SimilarityHook(nlp.vocab, **cfg),
'textcat': lambda nlp, **cfg: TextCategorizer(nlp.vocab, **cfg)
}
@ -150,7 +147,7 @@ class Language(object):
@property
def meta(self):
self._meta.setdefault('lang', self.vocab.lang)
self._meta.setdefault('name', '')
self._meta.setdefault('name', 'model')
self._meta.setdefault('version', '0.0.0')
self._meta.setdefault('spacy_version', about.__version__)
self._meta.setdefault('description', '')
@ -334,6 +331,29 @@ class Language(object):
doc = proc(doc)
return doc
def disable_pipes(self, *names):
'''Disable one or more pipeline components.
If used as a context manager, the pipeline will be restored to the initial
state at the end of the block. Otherwise, a DisabledPipes object is
returned, that has a `.restore()` method you can use to undo your
changes.
EXAMPLE:
>>> nlp.add_pipe('parser')
>>> nlp.add_pipe('tagger')
>>> with nlp.disable_pipes('parser', 'tagger'):
>>> assert not nlp.has_pipe('parser')
>>> assert nlp.has_pipe('parser')
>>> disabled = nlp.disable_pipes('parser')
>>> assert len(disabled) == 1
>>> assert not nlp.has_pipe('parser')
>>> disabled.restore()
>>> assert nlp.has_pipe('parser')
'''
return DisabledPipes(self, *names)
def make_doc(self, text):
return self.tokenizer(text)
@ -662,6 +682,42 @@ class Language(object):
return self
class DisabledPipes(list):
'''Manager for temporary pipeline disabling.'''
def __init__(self, nlp, *names):
self.nlp = nlp
self.names = names
# Important! Not deep copy -- we just want the container (but we also
# want to support people providing arbitrarily typed nlp.pipeline
# objects.)
self.original_pipeline = copy.copy(nlp.pipeline)
list.__init__(self)
self.extend(nlp.remove_pipe(name) for name in names)
def __enter__(self):
return self
def __exit__(self, *args):
self.restore()
def restore(self):
'''Restore the pipeline to its state when DisabledPipes was created.'''
current, self.nlp.pipeline = self.nlp.pipeline, self.original_pipeline
unexpected = [name for name, pipe in current if not self.nlp.has_pipe(name)]
if unexpected:
# Don't change the pipeline if we're raising an error.
self.nlp.pipeline = current
msg = (
"Some current components would be lost when restoring "
"previous pipeline state. If you added components after "
"calling nlp.disable_pipes(), you should remove them "
"explicitly with nlp.remove_pipe() before the pipeline is "
"restore. Names of the new components: %s"
)
raise ValueError(msg % unexpected)
self[:] = []
def unpickle_language(vocab, meta, bytes_data):
lang = Language(vocab=vocab)
lang.from_bytes(bytes_data)

21
spacy/pipeline.pxd
View File

@ -1,21 +0,0 @@
from .syntax.parser cimport Parser
#from .syntax.beam_parser cimport BeamParser
from .syntax.ner cimport BiluoPushDown
from .syntax.arc_eager cimport ArcEager
from .tagger cimport Tagger
cdef class EntityRecognizer(Parser):
pass
cdef class DependencyParser(Parser):
pass
#cdef class BeamEntityRecognizer(BeamParser):
# pass
#
#
#cdef class BeamDependencyParser(BeamParser):
# pass

94
spacy/pipeline.pyx
View File

@ -26,11 +26,8 @@ from thinc.neural.util import to_categorical
from thinc.neural._classes.difference import Siamese, CauchySimilarity
from .tokens.doc cimport Doc
from .syntax.parser cimport Parser as LinearParser
from .syntax.nn_parser cimport Parser as NeuralParser
from .syntax.nn_parser cimport Parser
from .syntax import nonproj
from .syntax.parser import get_templates as get_feature_templates
from .syntax.beam_parser cimport BeamParser
from .syntax.ner cimport BiluoPushDown
from .syntax.arc_eager cimport ArcEager
from .tagger import Tagger
@ -86,7 +83,7 @@ class SentenceSegmenter(object):
yield doc[start : len(doc)]
class BaseThincComponent(object):
class Pipe(object):
name = None
@classmethod
@ -217,7 +214,7 @@ def _load_cfg(path):
return {}
class TokenVectorEncoder(BaseThincComponent):
class Tensorizer(Pipe):
"""Assign position-sensitive vectors to tokens, using a CNN or RNN."""
name = 'tensorizer'
@ -329,7 +326,7 @@ class TokenVectorEncoder(BaseThincComponent):
link_vectors_to_models(self.vocab)
class NeuralTagger(BaseThincComponent):
class Tagger(Pipe):
name = 'tagger'
def __init__(self, vocab, model=True, **cfg):
self.vocab = vocab
@ -420,8 +417,6 @@ class NeuralTagger(BaseThincComponent):
new_tag_map[tag] = orig_tag_map[tag]
else:
new_tag_map[tag] = {POS: X}
if 'SP' not in new_tag_map:
new_tag_map['SP'] = orig_tag_map.get('SP', {POS: X})
cdef Vocab vocab = self.vocab
if new_tag_map:
vocab.morphology = Morphology(vocab.strings, new_tag_map,
@ -513,7 +508,11 @@ class NeuralTagger(BaseThincComponent):
return self
class NeuralLabeller(NeuralTagger):
class MultitaskObjective(Tagger):
'''Assist training of a parser or tagger, by training a side-objective.
Experimental
'''
name = 'nn_labeller'
def __init__(self, vocab, model=True, target='dep_tag_offset', **cfg):
self.vocab = vocab
@ -532,7 +531,7 @@ class NeuralLabeller(NeuralTagger):
self.make_label = target
else:
raise ValueError(
"NeuralLabeller target should be function or one of "
"MultitaskObjective target should be function or one of "
"['dep', 'tag', 'ent', 'dep_tag_offset', 'ent_tag']")
self.cfg = dict(cfg)
self.cfg.setdefault('cnn_maxout_pieces', 2)
@ -622,7 +621,7 @@ class NeuralLabeller(NeuralTagger):
return '%s-%s' % (tags[i], ents[i])
class SimilarityHook(BaseThincComponent):
class SimilarityHook(Pipe):
"""
Experimental
@ -674,7 +673,7 @@ class SimilarityHook(BaseThincComponent):
link_vectors_to_models(self.vocab)
class TextCategorizer(BaseThincComponent):
class TextCategorizer(Pipe):
name = 'textcat'
@classmethod
@ -752,45 +751,7 @@ class TextCategorizer(BaseThincComponent):
link_vectors_to_models(self.vocab)
cdef class EntityRecognizer(LinearParser):
"""Annotate named entities on Doc objects."""
TransitionSystem = BiluoPushDown
feature_templates = get_feature_templates('ner')
def add_label(self, label):
LinearParser.add_label(self, label)
if isinstance(label, basestring):
label = self.vocab.strings[label]
cdef class BeamEntityRecognizer(BeamParser):
"""Annotate named entities on Doc objects."""
TransitionSystem = BiluoPushDown
feature_templates = get_feature_templates('ner')
def add_label(self, label):
LinearParser.add_label(self, label)
if isinstance(label, basestring):
label = self.vocab.strings[label]
cdef class DependencyParser(LinearParser):
TransitionSystem = ArcEager
feature_templates = get_feature_templates('basic')
def add_label(self, label):
LinearParser.add_label(self, label)
if isinstance(label, basestring):
label = self.vocab.strings[label]
@property
def postprocesses(self):
return [nonproj.deprojectivize]
cdef class NeuralDependencyParser(NeuralParser):
cdef class DependencyParser(Parser):
name = 'parser'
TransitionSystem = ArcEager
@ -800,17 +761,17 @@ cdef class NeuralDependencyParser(NeuralParser):
def init_multitask_objectives(self, gold_tuples, pipeline, **cfg):
for target in []:
labeller = NeuralLabeller(self.vocab, target=target)
labeller = MultitaskObjective(self.vocab, target=target)
tok2vec = self.model[0]
labeller.begin_training(gold_tuples, pipeline=pipeline, tok2vec=tok2vec)
pipeline.append(labeller)
self._multitasks.append(labeller)
def __reduce__(self):
return (NeuralDependencyParser, (self.vocab, self.moves, self.model), None, None)
return (DependencyParser, (self.vocab, self.moves, self.model), None, None)
cdef class NeuralEntityRecognizer(NeuralParser):
cdef class EntityRecognizer(Parser):
name = 'ner'
TransitionSystem = BiluoPushDown
@ -818,31 +779,14 @@ cdef class NeuralEntityRecognizer(NeuralParser):
def init_multitask_objectives(self, gold_tuples, pipeline, **cfg):
for target in []:
labeller = NeuralLabeller(self.vocab, target=target)
labeller = MultitaskObjective(self.vocab, target=target)
tok2vec = self.model[0]
labeller.begin_training(gold_tuples, pipeline=pipeline, tok2vec=tok2vec)
pipeline.append(labeller)
self._multitasks.append(labeller)
def __reduce__(self):
return (NeuralEntityRecognizer, (self.vocab, self.moves, self.model), None, None)
return (EntityRecognizer, (self.vocab, self.moves, self.model), None, None)
cdef class BeamDependencyParser(BeamParser):
TransitionSystem = ArcEager
feature_templates = get_feature_templates('basic')
def add_label(self, label):
Parser.add_label(self, label)
if isinstance(label, basestring):
label = self.vocab.strings[label]
@property
def postprocesses(self):
return [nonproj.deprojectivize]
__all__ = ['Tagger', 'DependencyParser', 'EntityRecognizer', 'BeamDependencyParser',
'BeamEntityRecognizer', 'TokenVectorEnoder']
__all__ = ['Tagger', 'DependencyParser', 'EntityRecognizer', 'Tensorizer']

259
spacy/syntax/_parse_features.pxd
View File

@ -1,259 +0,0 @@
from thinc.typedefs cimport atom_t
from .stateclass cimport StateClass
from ._state cimport StateC
cdef int fill_context(atom_t* context, const StateC* state) nogil
# Context elements
# Ensure each token's attributes are listed: w, p, c, c6, c4. The order
# is referenced by incrementing the enum...
# Tokens are listed in left-to-right order.
#cdef size_t* SLOTS = [
# S2w, S1w,
# S0l0w, S0l2w, S0lw,
# S0w,
# S0r0w, S0r2w, S0rw,
# N0l0w, N0l2w, N0lw,
# P2w, P1w,
# N0w, N1w, N2w, N3w, 0
#]
# NB: The order of the enum is _NOT_ arbitrary!!
cpdef enum:
S2w
S2W
S2p
S2c
S2c4
S2c6
S2L
S2_prefix
S2_suffix
S2_shape
S2_ne_iob
S2_ne_type
S1w
S1W
S1p
S1c
S1c4
S1c6
S1L
S1_prefix
S1_suffix
S1_shape
S1_ne_iob
S1_ne_type
S1rw
S1rW
S1rp
S1rc
S1rc4
S1rc6
S1rL
S1r_prefix
S1r_suffix
S1r_shape
S1r_ne_iob
S1r_ne_type
S0lw
S0lW
S0lp
S0lc
S0lc4
S0lc6
S0lL
S0l_prefix
S0l_suffix
S0l_shape
S0l_ne_iob
S0l_ne_type
S0l2w
S0l2W
S0l2p
S0l2c
S0l2c4
S0l2c6
S0l2L
S0l2_prefix
S0l2_suffix
S0l2_shape
S0l2_ne_iob
S0l2_ne_type
S0w
S0W
S0p
S0c
S0c4
S0c6
S0L
S0_prefix
S0_suffix
S0_shape
S0_ne_iob
S0_ne_type
S0r2w
S0r2W
S0r2p
S0r2c
S0r2c4
S0r2c6
S0r2L
S0r2_prefix
S0r2_suffix
S0r2_shape
S0r2_ne_iob
S0r2_ne_type
S0rw
S0rW
S0rp
S0rc
S0rc4
S0rc6
S0rL
S0r_prefix
S0r_suffix
S0r_shape
S0r_ne_iob
S0r_ne_type
N0l2w
N0l2W
N0l2p
N0l2c
N0l2c4
N0l2c6
N0l2L
N0l2_prefix
N0l2_suffix
N0l2_shape
N0l2_ne_iob
N0l2_ne_type
N0lw
N0lW
N0lp
N0lc
N0lc4
N0lc6
N0lL
N0l_prefix
N0l_suffix
N0l_shape
N0l_ne_iob
N0l_ne_type
N0w
N0W
N0p
N0c
N0c4
N0c6
N0L
N0_prefix
N0_suffix
N0_shape
N0_ne_iob
N0_ne_type
N1w
N1W
N1p
N1c
N1c4
N1c6
N1L
N1_prefix
N1_suffix
N1_shape
N1_ne_iob
N1_ne_type
N2w
N2W
N2p
N2c
N2c4
N2c6
N2L
N2_prefix
N2_suffix
N2_shape
N2_ne_iob
N2_ne_type
P1w
P1W
P1p
P1c
P1c4
P1c6
P1L
P1_prefix
P1_suffix
P1_shape
P1_ne_iob
P1_ne_type
P2w
P2W
P2p
P2c
P2c4
P2c6
P2L
P2_prefix
P2_suffix
P2_shape
P2_ne_iob
P2_ne_type
E0w
E0W
E0p
E0c
E0c4
E0c6
E0L
E0_prefix
E0_suffix
E0_shape
E0_ne_iob
E0_ne_type
E1w
E1W
E1p
E1c
E1c4
E1c6
E1L
E1_prefix
E1_suffix
E1_shape
E1_ne_iob
E1_ne_type
# Misc features at the end
dist
N0lv
S0lv
S0rv
S1lv
S1rv
S0_has_head
S1_has_head
S2_has_head
CONTEXT_SIZE

419
spacy/syntax/_parse_features.pyx
View File

@ -1,419 +0,0 @@
"""
Fill an array, context, with every _atomic_ value our features reference.
We then write the _actual features_ as tuples of the atoms. The machinery
that translates from the tuples to feature-extractors (which pick the values
out of "context") is in features/extractor.pyx
The atomic feature names are listed in a big enum, so that the feature tuples
can refer to them.
"""
# coding: utf-8
from __future__ import unicode_literals
from libc.string cimport memset
from itertools import combinations
from cymem.cymem cimport Pool
from ..structs cimport TokenC
from .stateclass cimport StateClass
from ._state cimport StateC
cdef inline void fill_token(atom_t* context, const TokenC* token) nogil:
if token is NULL:
context[0] = 0
context[1] = 0
context[2] = 0
context[3] = 0
context[4] = 0
context[5] = 0
context[6] = 0
context[7] = 0
context[8] = 0
context[9] = 0
context[10] = 0
context[11] = 0
else:
context[0] = token.lex.orth
context[1] = token.lemma
context[2] = token.tag
context[3] = token.lex.cluster
# We've read in the string little-endian, so now we can take & (2**n)-1
# to get the first n bits of the cluster.
# e.g. s = "1110010101"
# s = ''.join(reversed(s))
# first_4_bits = int(s, 2)
# print first_4_bits
# 5
# print "{0:b}".format(prefix).ljust(4, '0')
# 1110
# What we're doing here is picking a number where all bits are 1, e.g.
# 15 is 1111, 63 is 111111 and doing bitwise AND, so getting all bits in
# the source that are set to 1.
context[4] = token.lex.cluster & 15
context[5] = token.lex.cluster & 63
context[6] = token.dep if token.head != 0 else 0
context[7] = token.lex.prefix
context[8] = token.lex.suffix
context[9] = token.lex.shape
context[10] = token.ent_iob
context[11] = token.ent_type
cdef int fill_context(atom_t* ctxt, const StateC* st) nogil:
# Take care to fill every element of context!
# We could memset, but this makes it very easy to have broken features that
# make almost no impact on accuracy. If instead they're unset, the impact
# tends to be dramatic, so we get an obvious regression to fix...
fill_token(&ctxt[S2w], st.S_(2))
fill_token(&ctxt[S1w], st.S_(1))
fill_token(&ctxt[S1rw], st.R_(st.S(1), 1))
fill_token(&ctxt[S0lw], st.L_(st.S(0), 1))
fill_token(&ctxt[S0l2w], st.L_(st.S(0), 2))
fill_token(&ctxt[S0w], st.S_(0))
fill_token(&ctxt[S0r2w], st.R_(st.S(0), 2))
fill_token(&ctxt[S0rw], st.R_(st.S(0), 1))
fill_token(&ctxt[N0lw], st.L_(st.B(0), 1))
fill_token(&ctxt[N0l2w], st.L_(st.B(0), 2))
fill_token(&ctxt[N0w], st.B_(0))
fill_token(&ctxt[N1w], st.B_(1))
fill_token(&ctxt[N2w], st.B_(2))
fill_token(&ctxt[P1w], st.safe_get(st.B(0)-1))
fill_token(&ctxt[P2w], st.safe_get(st.B(0)-2))
fill_token(&ctxt[E0w], st.E_(0))
fill_token(&ctxt[E1w], st.E_(1))
if st.stack_depth() >= 1 and not st.eol():
ctxt[dist] = min_(st.B(0) - st.E(0), 5)
else:
ctxt[dist] = 0
ctxt[N0lv] = min_(st.n_L(st.B(0)), 5)
ctxt[S0lv] = min_(st.n_L(st.S(0)), 5)
ctxt[S0rv] = min_(st.n_R(st.S(0)), 5)
ctxt[S1lv] = min_(st.n_L(st.S(1)), 5)
ctxt[S1rv] = min_(st.n_R(st.S(1)), 5)
ctxt[S0_has_head] = 0
ctxt[S1_has_head] = 0
ctxt[S2_has_head] = 0
if st.stack_depth() >= 1:
ctxt[S0_has_head] = st.has_head(st.S(0)) + 1
if st.stack_depth() >= 2:
ctxt[S1_has_head] = st.has_head(st.S(1)) + 1
if st.stack_depth() >= 3:
ctxt[S2_has_head] = st.has_head(st.S(2)) + 1
cdef inline int min_(int a, int b) nogil:
return a if a > b else b
ner = (
(N0W,),
(P1W,),
(N1W,),
(P2W,),
(N2W,),
(P1W, N0W,),
(N0W, N1W),
(N0_prefix,),
(N0_suffix,),
(P1_shape,),
(N0_shape,),
(N1_shape,),
(P1_shape, N0_shape,),
(N0_shape, P1_shape,),
(P1_shape, N0_shape, N1_shape),
(N2_shape,),
(P2_shape,),
#(P2_norm, P1_norm, W_norm),
#(P1_norm, W_norm, N1_norm),
#(W_norm, N1_norm, N2_norm)
(P2p,),
(P1p,),
(N0p,),
(N1p,),
(N2p,),
(P1p, N0p),
(N0p, N1p),
(P2p, P1p, N0p),
(P1p, N0p, N1p),
(N0p, N1p, N2p),
(P2c,),
(P1c,),
(N0c,),
(N1c,),
(N2c,),
(P1c, N0c),
(N0c, N1c),
(E0W,),
(E0c,),
(E0p,),
(E0W, N0W),
(E0c, N0W),
(E0p, N0W),
(E0p, P1p, N0p),
(E0c, P1c, N0c),
(E0w, P1c),
(E0p, P1p),
(E0c, P1c),
(E0p, E1p),
(E0c, P1p),
(E1W,),
(E1c,),
(E1p,),
(E0W, E1W),
(E0W, E1p,),
(E0p, E1W,),
(E0p, E1W),
(P1_ne_iob,),
(P1_ne_iob, P1_ne_type),
(N0w, P1_ne_iob, P1_ne_type),
(N0_shape,),
(N1_shape,),
(N2_shape,),
(P1_shape,),
(P2_shape,),
(N0_prefix,),
(N0_suffix,),
(P1_ne_iob,),
(P2_ne_iob,),
(P1_ne_iob, P2_ne_iob),
(P1_ne_iob, P1_ne_type),
(P2_ne_iob, P2_ne_type),
(N0w, P1_ne_iob, P1_ne_type),
(N0w, N1w),
)
unigrams = (
(S2W, S2p),
(S2c6, S2p),
(S1W, S1p),
(S1c6, S1p),
(S0W, S0p),
(S0c6, S0p),
(N0W, N0p),
(N0p,),
(N0c,),
(N0c6, N0p),
(N0L,),
(N1W, N1p),
(N1c6, N1p),
(N2W, N2p),
(N2c6, N2p),
(S0r2W, S0r2p),
(S0r2c6, S0r2p),
(S0r2L,),
(S0rW, S0rp),
(S0rc6, S0rp),
(S0rL,),
(S0l2W, S0l2p),
(S0l2c6, S0l2p),
(S0l2L,),
(S0lW, S0lp),
(S0lc6, S0lp),
(S0lL,),
(N0l2W, N0l2p),
(N0l2c6, N0l2p),
(N0l2L,),
(N0lW, N0lp),
(N0lc6, N0lp),
(N0lL,),
)
s0_n0 = (
(S0W, S0p, N0W, N0p),
(S0c, S0p, N0c, N0p),
(S0c6, S0p, N0c6, N0p),
(S0c4, S0p, N0c4, N0p),
(S0p, N0p),
(S0W, N0p),
(S0p, N0W),
(S0W, N0c),
(S0c, N0W),
(S0p, N0c),
(S0c, N0p),
(S0W, S0rp, N0p),
(S0p, S0rp, N0p),
(S0p, N0lp, N0W),
(S0p, N0lp, N0p),
(S0L, N0p),
(S0p, S0rL, N0p),
(S0p, N0lL, N0p),
(S0p, S0rv, N0p),
(S0p, N0lv, N0p),
(S0c6, S0rL, S0r2L, N0p),
(S0p, N0lL, N0l2L, N0p),
)
s1_s0 = (
(S1p, S0p),
(S1p, S0p, S0_has_head),
(S1W, S0p),
(S1W, S0p, S0_has_head),
(S1c, S0p),
(S1c, S0p, S0_has_head),
(S1p, S1rL, S0p),
(S1p, S1rL, S0p, S0_has_head),
(S1p, S0lL, S0p),
(S1p, S0lL, S0p, S0_has_head),
(S1p, S0lL, S0l2L, S0p),
(S1p, S0lL, S0l2L, S0p, S0_has_head),
(S1L, S0L, S0W),
(S1L, S0L, S0p),
(S1p, S1L, S0L, S0p),
(S1p, S0p),
)
s1_n0 = (
(S1p, N0p),
(S1c, N0c),
(S1c, N0p),
(S1p, N0c),
(S1W, S1p, N0p),
(S1p, N0W, N0p),
(S1c6, S1p, N0c6, N0p),
(S1L, N0p),
(S1p, S1rL, N0p),
(S1p, S1rp, N0p),
)
s0_n1 = (
(S0p, N1p),
(S0c, N1c),
(S0c, N1p),
(S0p, N1c),
(S0W, S0p, N1p),
(S0p, N1W, N1p),
(S0c6, S0p, N1c6, N1p),
(S0L, N1p),
(S0p, S0rL, N1p),
)
n0_n1 = (
(N0W, N0p, N1W, N1p),
(N0W, N0p, N1p),
(N0p, N1W, N1p),
(N0c, N0p, N1c, N1p),
(N0c6, N0p, N1c6, N1p),
(N0c, N1c),
(N0p, N1c),
)
tree_shape = (
(dist,),
(S0p, S0_has_head, S1_has_head, S2_has_head),
(S0p, S0lv, S0rv),
(N0p, N0lv),
)
trigrams = (
(N0p, N1p, N2p),
(S0p, S0lp, S0l2p),
(S0p, S0rp, S0r2p),
(S0p, S1p, S2p),
(S1p, S0p, N0p),
(S0p, S0lp, N0p),
(S0p, N0p, N0lp),
(N0p, N0lp, N0l2p),
(S0W, S0p, S0rL, S0r2L),
(S0p, S0rL, S0r2L),
(S0W, S0p, S0lL, S0l2L),
(S0p, S0lL, S0l2L),
(N0W, N0p, N0lL, N0l2L),
(N0p, N0lL, N0l2L),
)
words = (
S2w,
S1w,
S1rw,
S0lw,
S0l2w,
S0w,
S0r2w,
S0rw,
N0lw,
N0l2w,
N0w,
N1w,
N2w,
P1w,
P2w
)
tags = (
S2p,
S1p,
S1rp,
S0lp,
S0l2p,
S0p,
S0r2p,
S0rp,
N0lp,
N0l2p,
N0p,
N1p,
N2p,
P1p,
P2p
)
labels = (
S2L,
S1L,
S1rL,
S0lL,
S0l2L,
S0L,
S0r2L,
S0rL,
N0lL,
N0l2L,
N0L,
N1L,
N2L,
P1L,
P2L
)

10
spacy/syntax/beam_parser.pxd
View File

@ -1,10 +0,0 @@
from .parser cimport Parser
from ..structs cimport TokenC
from thinc.typedefs cimport weight_t
cdef class BeamParser(Parser):
cdef public int beam_width
cdef public weight_t beam_density
cdef int _parseC(self, TokenC* tokens, int length, int nr_feat, int nr_class) except -1

239
spacy/syntax/beam_parser.pyx
View File

@ -1,239 +0,0 @@
"""
MALT-style dependency parser
"""
# cython: profile=True
# cython: experimental_cpp_class_def=True
# cython: cdivision=True
# cython: infer_types=True
# coding: utf-8
from __future__ import unicode_literals, print_function
cimport cython
from cpython.ref cimport PyObject, Py_INCREF, Py_XDECREF
from libc.stdint cimport uint32_t, uint64_t
from libc.string cimport memset, memcpy
from libc.stdlib cimport rand
from libc.math cimport log, exp, isnan, isinf
from cymem.cymem cimport Pool, Address
from murmurhash.mrmr cimport real_hash64 as hash64
from thinc.typedefs cimport weight_t, class_t, feat_t, atom_t, hash_t
from thinc.linear.features cimport ConjunctionExtracter
from thinc.structs cimport FeatureC, ExampleC
from thinc.extra.search cimport Beam, MaxViolation
from thinc.extra.eg cimport Example
from thinc.extra.mb cimport Minibatch
from ..structs cimport TokenC
from ..tokens.doc cimport Doc
from ..strings cimport StringStore
from .transition_system cimport TransitionSystem, Transition
from ..gold cimport GoldParse
from . import _parse_features
from ._parse_features cimport CONTEXT_SIZE
from ._parse_features cimport fill_context
from .stateclass cimport StateClass
from .parser cimport Parser
DEBUG = False
def set_debug(val):
global DEBUG
DEBUG = val
def get_templates(name):
pf = _parse_features
if name == 'ner':
return pf.ner
elif name == 'debug':
return pf.unigrams
else:
return (pf.unigrams + pf.s0_n0 + pf.s1_n0 + pf.s1_s0 + pf.s0_n1 + pf.n0_n1 + \
pf.tree_shape + pf.trigrams)
cdef int BEAM_WIDTH = 16
cdef weight_t BEAM_DENSITY = 0.001
cdef class BeamParser(Parser):
def __init__(self, *args, **kwargs):
self.beam_width = kwargs.get('beam_width', BEAM_WIDTH)
self.beam_density = kwargs.get('beam_density', BEAM_DENSITY)
Parser.__init__(self, *args, **kwargs)
cdef int parseC(self, TokenC* tokens, int length, int nr_feat) nogil:
with gil:
self._parseC(tokens, length, nr_feat, self.moves.n_moves)
cdef int _parseC(self, TokenC* tokens, int length, int nr_feat, int nr_class) except -1:
cdef Beam beam = Beam(self.moves.n_moves, self.beam_width, min_density=self.beam_density)
# TODO: How do we handle new labels here? This increases nr_class
beam.initialize(self.moves.init_beam_state, length, tokens)
beam.check_done(_check_final_state, NULL)
if beam.is_done:
_cleanup(beam)
return 0
while not beam.is_done:
self._advance_beam(beam, None, False)
state = <StateClass>beam.at(0)
self.moves.finalize_state(state.c)
for i in range(length):
tokens[i] = state.c._sent[i]
_cleanup(beam)
def update(self, Doc tokens, GoldParse gold_parse, itn=0):
self.moves.preprocess_gold(gold_parse)
cdef Beam pred = Beam(self.moves.n_moves, self.beam_width)
pred.initialize(self.moves.init_beam_state, tokens.length, tokens.c)
pred.check_done(_check_final_state, NULL)
# Hack for NER
for i in range(pred.size):
stcls = <StateClass>pred.at(i)
self.moves.initialize_state(stcls.c)
cdef Beam gold = Beam(self.moves.n_moves, self.beam_width, min_density=0.0)
gold.initialize(self.moves.init_beam_state, tokens.length, tokens.c)
gold.check_done(_check_final_state, NULL)
violn = MaxViolation()
while not pred.is_done and not gold.is_done:
# We search separately here, to allow for ambiguity in the gold parse.
self._advance_beam(pred, gold_parse, False)
self._advance_beam(gold, gold_parse, True)
violn.check_crf(pred, gold)
if pred.loss > 0 and pred.min_score > (gold.score + self.model.time):
break
else:
# The non-monotonic oracle makes it difficult to ensure final costs are
# correct. Therefore do final correction
for i in range(pred.size):
if self.moves.is_gold_parse(<StateClass>pred.at(i), gold_parse):
pred._states[i].loss = 0.0
elif pred._states[i].loss == 0.0:
pred._states[i].loss = 1.0
violn.check_crf(pred, gold)
if pred.size < 1:
raise Exception("No candidates", tokens.length)
if gold.size < 1:
raise Exception("No gold", tokens.length)
if pred.loss == 0:
self.model.update_from_histories(self.moves, tokens, [(0.0, [])])
elif True:
#_check_train_integrity(pred, gold, gold_parse, self.moves)
histories = list(zip(violn.p_probs, violn.p_hist)) + \
list(zip(violn.g_probs, violn.g_hist))
self.model.update_from_histories(self.moves, tokens, histories, min_grad=0.001**(itn+1))
else:
self.model.update_from_histories(self.moves, tokens,
[(1.0, violn.p_hist[0]), (-1.0, violn.g_hist[0])])
_cleanup(pred)
_cleanup(gold)
return pred.loss
def _advance_beam(self, Beam beam, GoldParse gold, bint follow_gold):
cdef atom_t[CONTEXT_SIZE] context
cdef Pool mem = Pool()
features = <FeatureC*>mem.alloc(self.model.nr_feat, sizeof(FeatureC))
if False:
mb = Minibatch(self.model.widths, beam.size)
for i in range(beam.size):
stcls = <StateClass>beam.at(i)
if stcls.c.is_final():
nr_feat = 0
else:
nr_feat = self.model.set_featuresC(context, features, stcls.c)
self.moves.set_valid(beam.is_valid[i], stcls.c)
mb.c.push_back(features, nr_feat, beam.costs[i], beam.is_valid[i], 0)
self.model(mb)
for i in range(beam.size):
memcpy(beam.scores[i], mb.c.scores(i), mb.c.nr_out() * sizeof(beam.scores[i][0]))
else:
for i in range(beam.size):
stcls = <StateClass>beam.at(i)
if not stcls.is_final():
nr_feat = self.model.set_featuresC(context, features, stcls.c)
self.moves.set_valid(beam.is_valid[i], stcls.c)
self.model.set_scoresC(beam.scores[i], features, nr_feat)
if gold is not None:
n_gold = 0
lines = []
for i in range(beam.size):
stcls = <StateClass>beam.at(i)
if not stcls.c.is_final():
self.moves.set_costs(beam.is_valid[i], beam.costs[i], stcls, gold)
if follow_gold:
for j in range(self.moves.n_moves):
if beam.costs[i][j] >= 1:
beam.is_valid[i][j] = 0
lines.append((stcls.B(0), stcls.B(1),
stcls.B_(0).ent_iob, stcls.B_(1).ent_iob,
stcls.B_(1).sent_start,
j,
beam.is_valid[i][j], 'set invalid',
beam.costs[i][j], self.moves.c[j].move, self.moves.c[j].label))
n_gold += 1 if beam.is_valid[i][j] else 0
if follow_gold and n_gold == 0:
raise Exception("No gold")
if follow_gold:
beam.advance(_transition_state, NULL, <void*>self.moves.c)
else:
beam.advance(_transition_state, _hash_state, <void*>self.moves.c)
beam.check_done(_check_final_state, NULL)
# These are passed as callbacks to thinc.search.Beam
cdef int _transition_state(void* _dest, void* _src, class_t clas, void* _moves) except -1:
dest = <StateClass>_dest
src = <StateClass>_src
moves = <const Transition*>_moves
dest.clone(src)
moves[clas].do(dest.c, moves[clas].label)
cdef int _check_final_state(void* _state, void* extra_args) except -1:
return (<StateClass>_state).is_final()
def _cleanup(Beam beam):
for i in range(beam.width):
Py_XDECREF(<PyObject*>beam._states[i].content)
Py_XDECREF(<PyObject*>beam._parents[i].content)
cdef hash_t _hash_state(void* _state, void* _) except 0:
state = <StateClass>_state
if state.c.is_final():
return 1
else:
return state.c.hash()
def _check_train_integrity(Beam pred, Beam gold, GoldParse gold_parse, TransitionSystem moves):
for i in range(pred.size):
if not pred._states[i].is_done or pred._states[i].loss == 0:
continue
state = <StateClass>pred.at(i)
if moves.is_gold_parse(state, gold_parse) == True:
for dep in gold_parse.orig_annot:
print(dep[1], dep[3], dep[4])
print("Cost", pred._states[i].loss)
for j in range(gold_parse.length):
print(gold_parse.orig_annot[j][1], state.H(j), moves.strings[state.safe_get(j).dep])
acts = [moves.c[clas].move for clas in pred.histories[i]]
labels = [moves.c[clas].label for clas in pred.histories[i]]
print([moves.move_name(move, label) for move, label in zip(acts, labels)])
raise Exception("Predicted state is gold-standard")
for i in range(gold.size):
if not gold._states[i].is_done:
continue
state = <StateClass>gold.at(i)
if moves.is_gold(state, gold_parse) == False:
print("Truth")
for dep in gold_parse.orig_annot:
print(dep[1], dep[3], dep[4])
print("Predicted good")
for j in range(gold_parse.length):
print(gold_parse.orig_annot[j][1], state.H(j), moves.strings[state.safe_get(j).dep])
raise Exception("Gold parse is not gold-standard")

3
spacy/syntax/nn_parser.pyx
View File

@ -53,9 +53,6 @@ from .._ml import link_vectors_to_models
from .._ml import HistoryFeatures
from ..compat import json_dumps, copy_array
from . import _parse_features
from ._parse_features cimport CONTEXT_SIZE
from ._parse_features cimport fill_context
from .stateclass cimport StateClass
from ._state cimport StateC
from . import nonproj

24
spacy/syntax/parser.pxd
View File

@ -1,24 +0,0 @@
from thinc.linear.avgtron cimport AveragedPerceptron
from thinc.typedefs cimport atom_t
from thinc.structs cimport FeatureC
from .stateclass cimport StateClass
from .arc_eager cimport TransitionSystem
from ..vocab cimport Vocab
from ..tokens.doc cimport Doc
from ..structs cimport TokenC
from ._state cimport StateC
cdef class ParserModel(AveragedPerceptron):
cdef int set_featuresC(self, atom_t* context, FeatureC* features,
const StateC* state) nogil
cdef class Parser:
cdef readonly Vocab vocab
cdef readonly ParserModel model
cdef readonly TransitionSystem moves
cdef readonly object cfg
cdef int parseC(self, TokenC* tokens, int length, int nr_feat) nogil

526
spacy/syntax/parser.pyx
View File

@ -1,526 +0,0 @@
"""
MALT-style dependency parser
"""
# coding: utf-8
# cython: infer_types=True
from __future__ import unicode_literals
from collections import Counter
import ujson
cimport cython
cimport cython.parallel
import numpy.random
from cpython.ref cimport PyObject, Py_INCREF, Py_XDECREF
from cpython.exc cimport PyErr_CheckSignals
from libc.stdint cimport uint32_t, uint64_t
from libc.string cimport memset, memcpy
from libc.stdlib cimport malloc, calloc, free
from thinc.typedefs cimport weight_t, class_t, feat_t, atom_t, hash_t
from thinc.linear.avgtron cimport AveragedPerceptron
from thinc.linalg cimport VecVec
from thinc.structs cimport SparseArrayC, FeatureC, ExampleC
from thinc.extra.eg cimport Example
from cymem.cymem cimport Pool, Address
from murmurhash.mrmr cimport hash64
from preshed.maps cimport MapStruct
from preshed.maps cimport map_get
from . import _parse_features
from ._parse_features cimport CONTEXT_SIZE
from ._parse_features cimport fill_context
from .stateclass cimport StateClass
from ._state cimport StateC
from .transition_system import OracleError
from .transition_system cimport TransitionSystem, Transition
from ..structs cimport TokenC
from ..tokens.doc cimport Doc
from ..strings cimport StringStore
from ..gold cimport GoldParse
USE_FTRL = True
DEBUG = False
def set_debug(val):
global DEBUG
DEBUG = val
def get_templates(name):
pf = _parse_features
if name == 'ner':
return pf.ner
elif name == 'debug':
return pf.unigrams
elif name.startswith('embed'):
return (pf.words, pf.tags, pf.labels)
else:
return (pf.unigrams + pf.s0_n0 + pf.s1_n0 + pf.s1_s0 + pf.s0_n1 + pf.n0_n1 + \
pf.tree_shape + pf.trigrams)
cdef class ParserModel(AveragedPerceptron):
cdef int set_featuresC(self, atom_t* context, FeatureC* features,
const StateC* state) nogil:
fill_context(context, state)
nr_feat = self.extracter.set_features(features, context)
return nr_feat
def update(self, Example eg, itn=0):
"""
Does regression on negative cost. Sort of cute?
"""
self.time += 1
cdef int best = arg_max_if_gold(eg.c.scores, eg.c.costs, eg.c.nr_class)
cdef int guess = eg.guess
if guess == best or best == -1:
return 0.0
cdef FeatureC feat
cdef int clas
cdef weight_t gradient
if USE_FTRL:
for feat in eg.c.features[:eg.c.nr_feat]:
for clas in range(eg.c.nr_class):
if eg.c.is_valid[clas] and eg.c.scores[clas] >= eg.c.scores[best]:
gradient = eg.c.scores[clas] + eg.c.costs[clas]
self.update_weight_ftrl(feat.key, clas, feat.value * gradient)
else:
for feat in eg.c.features[:eg.c.nr_feat]:
self.update_weight(feat.key, guess, feat.value * eg.c.costs[guess])
self.update_weight(feat.key, best, -feat.value * eg.c.costs[guess])
return eg.c.costs[guess]
def update_from_histories(self, TransitionSystem moves, Doc doc, histories, weight_t min_grad=0.0):
cdef Pool mem = Pool()
features = <FeatureC*>mem.alloc(self.nr_feat, sizeof(FeatureC))
cdef StateClass stcls
cdef class_t clas
self.time += 1
cdef atom_t[CONTEXT_SIZE] atoms
histories = [(grad, hist) for grad, hist in histories if abs(grad) >= min_grad and hist]
if not histories:
return None
gradient = [Counter() for _ in range(max([max(h)+1 for _, h in histories]))]
for d_loss, history in histories:
stcls = StateClass.init(doc.c, doc.length)
moves.initialize_state(stcls.c)
for clas in history:
nr_feat = self.set_featuresC(atoms, features, stcls.c)
clas_grad = gradient[clas]
for feat in features[:nr_feat]:
clas_grad[feat.key] += d_loss * feat.value
moves.c[clas].do(stcls.c, moves.c[clas].label)
cdef feat_t key
cdef weight_t d_feat
for clas, clas_grad in enumerate(gradient):
for key, d_feat in clas_grad.items():
if d_feat != 0:
self.update_weight_ftrl(key, clas, d_feat)
cdef class Parser:
"""
Base class of the DependencyParser and EntityRecognizer.
"""
@classmethod
def load(cls, path, Vocab vocab, TransitionSystem=None, require=False, **cfg):
"""
Load the statistical model from the supplied path.
Arguments:
path (Path):
The path to load from.
vocab (Vocab):
The vocabulary. Must be shared by the documents to be processed.
require (bool):
Whether to raise an error if the files are not found.
Returns (Parser):
The newly constructed object.
"""
with (path / 'config.json').open() as file_:
cfg = ujson.load(file_)
# TODO: remove this shim when we don't have to support older data
if 'labels' in cfg and 'actions' not in cfg:
cfg['actions'] = cfg.pop('labels')
# TODO: remove this shim when we don't have to support older data
for action_name, labels in dict(cfg.get('actions', {})).items():
# We need this to be sorted
if isinstance(labels, dict):
labels = list(sorted(labels.keys()))
cfg['actions'][action_name] = labels
self = cls(vocab, TransitionSystem=TransitionSystem, model=None, **cfg)
if (path / 'model').exists():
self.model.load(str(path / 'model'))
elif require:
raise IOError(
"Required file %s/model not found when loading" % str(path))
return self
def __init__(self, Vocab vocab, TransitionSystem=None, ParserModel model=None, **cfg):
"""
Create a Parser.
Arguments:
vocab (Vocab):
The vocabulary object. Must be shared with documents to be processed.
model (thinc.linear.AveragedPerceptron):
The statistical model.
Returns (Parser):
The newly constructed object.
"""
if TransitionSystem is None:
TransitionSystem = self.TransitionSystem
self.vocab = vocab
cfg['actions'] = TransitionSystem.get_actions(**cfg)
self.moves = TransitionSystem(vocab.strings, cfg['actions'])
# TODO: Remove this when we no longer need to support old-style models
if isinstance(cfg.get('features'), basestring):
cfg['features'] = get_templates(cfg['features'])
elif 'features' not in cfg:
cfg['features'] = self.feature_templates
self.model = ParserModel(cfg['features'])
self.model.l1_penalty = cfg.get('L1', 0.0)
self.model.learn_rate = cfg.get('learn_rate', 0.001)
self.cfg = cfg
# TODO: This is a pretty hacky fix to the problem of adding more
# labels. The issue is they come in out of order, if labels are
# added during training
for label in cfg.get('extra_labels', []):
self.add_label(label)
def __reduce__(self):
return (Parser, (self.vocab, self.moves, self.model), None, None)
def __call__(self, Doc tokens):
"""
Apply the entity recognizer, setting the annotations onto the Doc object.
Arguments:
doc (Doc): The document to be processed.
Returns:
None
"""
cdef int nr_feat = self.model.nr_feat
with nogil:
status = self.parseC(tokens.c, tokens.length, nr_feat)
# Check for KeyboardInterrupt etc. Untested
PyErr_CheckSignals()
if status != 0:
raise ParserStateError(tokens)
self.moves.finalize_doc(tokens)
def pipe(self, stream, int batch_size=1000, int n_threads=2):
"""
Process a stream of documents.
Arguments:
stream: The sequence of documents to process.
batch_size (int):
The number of documents to accumulate into a working set.
n_threads (int):
The number of threads with which to work on the buffer in parallel.
Yields (Doc): Documents, in order.
"""
cdef Pool mem = Pool()
cdef TokenC** doc_ptr = <TokenC**>mem.alloc(batch_size, sizeof(TokenC*))
cdef int* lengths = <int*>mem.alloc(batch_size, sizeof(int))
cdef Doc doc
cdef int i
cdef int nr_feat = self.model.nr_feat
cdef int status
queue = []
for doc in stream:
doc_ptr[len(queue)] = doc.c
lengths[len(queue)] = doc.length
queue.append(doc)
if len(queue) == batch_size:
with nogil:
for i in cython.parallel.prange(batch_size, num_threads=n_threads):
status = self.parseC(doc_ptr[i], lengths[i], nr_feat)
if status != 0:
with gil:
raise ParserStateError(queue[i])
PyErr_CheckSignals()
for doc in queue:
self.moves.finalize_doc(doc)
yield doc
queue = []
batch_size = len(queue)
with nogil:
for i in cython.parallel.prange(batch_size, num_threads=n_threads):
status = self.parseC(doc_ptr[i], lengths[i], nr_feat)
if status != 0:
with gil:
raise ParserStateError(queue[i])
PyErr_CheckSignals()
for doc in queue:
self.moves.finalize_doc(doc)
yield doc
cdef int parseC(self, TokenC* tokens, int length, int nr_feat) nogil:
state = new StateC(tokens, length)
# NB: This can change self.moves.n_moves!
# I think this causes memory errors if called by .pipe()
self.moves.initialize_state(state)
nr_class = self.moves.n_moves
cdef ExampleC eg
eg.nr_feat = nr_feat
eg.nr_atom = CONTEXT_SIZE
eg.nr_class = nr_class
eg.features = <FeatureC*>calloc(sizeof(FeatureC), nr_feat)
eg.atoms = <atom_t*>calloc(sizeof(atom_t), CONTEXT_SIZE)
eg.scores = <weight_t*>calloc(sizeof(weight_t), nr_class)
eg.is_valid = <int*>calloc(sizeof(int), nr_class)
cdef int i
while not state.is_final():
eg.nr_feat = self.model.set_featuresC(eg.atoms, eg.features, state)
self.moves.set_valid(eg.is_valid, state)
self.model.set_scoresC(eg.scores, eg.features, eg.nr_feat)
guess = VecVec.arg_max_if_true(eg.scores, eg.is_valid, eg.nr_class)
if guess < 0:
return 1
action = self.moves.c[guess]
action.do(state, action.label)
memset(eg.scores, 0, sizeof(eg.scores[0]) * eg.nr_class)
for i in range(eg.nr_class):
eg.is_valid[i] = 1
self.moves.finalize_state(state)
for i in range(length):
tokens[i] = state._sent[i]
del state
free(eg.features)
free(eg.atoms)
free(eg.scores)
free(eg.is_valid)
return 0
def update(self, Doc tokens, GoldParse gold, itn=0, double drop=0.0):
"""
Update the statistical model.
Arguments:
doc (Doc):
The example document for the update.
gold (GoldParse):
The gold-standard annotations, to calculate the loss.
Returns (float):
The loss on this example.
"""
self.moves.preprocess_gold(gold)
cdef StateClass stcls = StateClass.init(tokens.c, tokens.length)
self.moves.initialize_state(stcls.c)
cdef Pool mem = Pool()
cdef Example eg = Example(
nr_class=self.moves.n_moves,
nr_atom=CONTEXT_SIZE,
nr_feat=self.model.nr_feat)
cdef weight_t loss = 0
cdef Transition action
cdef double dropout_rate = self.cfg.get('dropout', drop)
while not stcls.is_final():
eg.c.nr_feat = self.model.set_featuresC(eg.c.atoms, eg.c.features,
stcls.c)
dropout(eg.c.features, eg.c.nr_feat, dropout_rate)
self.moves.set_costs(eg.c.is_valid, eg.c.costs, stcls, gold)
self.model.set_scoresC(eg.c.scores, eg.c.features, eg.c.nr_feat)
guess = VecVec.arg_max_if_true(eg.c.scores, eg.c.is_valid, eg.c.nr_class)
self.model.update(eg)
action = self.moves.c[guess]
action.do(stcls.c, action.label)
loss += eg.costs[guess]
eg.fill_scores(0, eg.c.nr_class)
eg.fill_costs(0, eg.c.nr_class)
eg.fill_is_valid(1, eg.c.nr_class)
self.moves.finalize_state(stcls.c)
return loss
def step_through(self, Doc doc, GoldParse gold=None):
"""
Set up a stepwise state, to introspect and control the transition sequence.
Arguments:
doc (Doc): The document to step through.
gold (GoldParse): Optional gold parse
Returns (StepwiseState):
A state object, to step through the annotation process.
"""
return StepwiseState(self, doc, gold=gold)
def from_transition_sequence(self, Doc doc, sequence):
"""Control the annotations on a document by specifying a transition sequence
to follow.
Arguments:
doc (Doc): The document to annotate.
sequence: A sequence of action names, as unicode strings.
Returns: None
"""
with self.step_through(doc) as stepwise:
for transition in sequence:
stepwise.transition(transition)
def add_label(self, label):
# Doesn't set label into serializer -- subclasses override it to do that.
for action in self.moves.action_types:
added = self.moves.add_action(action, label)
if added:
# Important that the labels be stored as a list! We need the
# order, or the model goes out of synch
self.cfg.setdefault('extra_labels', []).append(label)
cdef int dropout(FeatureC* feats, int nr_feat, float prob) except -1:
if prob <= 0 or prob >= 1.:
return 0
cdef double[::1] py_probs = numpy.random.uniform(0., 1., nr_feat)
cdef double* probs = &py_probs[0]
for i in range(nr_feat):
if probs[i] >= prob:
feats[i].value /= prob
else:
feats[i].value = 0.
cdef class StepwiseState:
cdef readonly StateClass stcls
cdef readonly Example eg
cdef readonly Doc doc
cdef readonly GoldParse gold
cdef readonly Parser parser
def __init__(self, Parser parser, Doc doc, GoldParse gold=None):
self.parser = parser
self.doc = doc
if gold is not None:
self.gold = gold
self.parser.moves.preprocess_gold(self.gold)
else:
self.gold = GoldParse(doc)
self.stcls = StateClass.init(doc.c, doc.length)
self.parser.moves.initialize_state(self.stcls.c)
self.eg = Example(
nr_class=self.parser.moves.n_moves,
nr_atom=CONTEXT_SIZE,
nr_feat=self.parser.model.nr_feat)
def __enter__(self):
return self
def __exit__(self, type, value, traceback):
self.finish()
@property
def is_final(self):
return self.stcls.is_final()
@property
def stack(self):
return self.stcls.stack
@property
def queue(self):
return self.stcls.queue
@property
def heads(self):
return [self.stcls.H(i) for i in range(self.stcls.c.length)]
@property
def deps(self):
return [self.doc.vocab.strings[self.stcls.c._sent[i].dep]
for i in range(self.stcls.c.length)]
@property
def costs(self):
"""
Find the action-costs for the current state.
"""
if not self.gold:
raise ValueError("Can't set costs: No GoldParse provided")
self.parser.moves.set_costs(self.eg.c.is_valid, self.eg.c.costs,
self.stcls, self.gold)
costs = {}
for i in range(self.parser.moves.n_moves):
if not self.eg.c.is_valid[i]:
continue
transition = self.parser.moves.c[i]
name = self.parser.moves.move_name(transition.move, transition.label)
costs[name] = self.eg.c.costs[i]
return costs
def predict(self):
self.eg.reset()
self.eg.c.nr_feat = self.parser.model.set_featuresC(self.eg.c.atoms, self.eg.c.features,
self.stcls.c)
self.parser.moves.set_valid(self.eg.c.is_valid, self.stcls.c)
self.parser.model.set_scoresC(self.eg.c.scores,
self.eg.c.features, self.eg.c.nr_feat)
cdef Transition action = self.parser.moves.c[self.eg.guess]
return self.parser.moves.move_name(action.move, action.label)
def transition(self, action_name=None):
if action_name is None:
action_name = self.predict()
moves = {'S': 0, 'D': 1, 'L': 2, 'R': 3}
if action_name == '_':
action_name = self.predict()
action = self.parser.moves.lookup_transition(action_name)
elif action_name == 'L' or action_name == 'R':
self.predict()
move = moves[action_name]
clas = _arg_max_clas(self.eg.c.scores, move, self.parser.moves.c,
self.eg.c.nr_class)
action = self.parser.moves.c[clas]
else:
action = self.parser.moves.lookup_transition(action_name)
action.do(self.stcls.c, action.label)
def finish(self):
if self.stcls.is_final():
self.parser.moves.finalize_state(self.stcls.c)
self.doc.set_parse(self.stcls.c._sent)
self.parser.moves.finalize_doc(self.doc)
class ParserStateError(ValueError):
def __init__(self, doc):
ValueError.__init__(self,
"Error analysing doc -- no valid actions available. This should "
"never happen, so please report the error on the issue tracker. "
"Here's the thread to do so --- reopen it if it's closed:\n"
"https://github.com/spacy-io/spaCy/issues/429\n"
"Please include the text that the parser failed on, which is:\n"
"%s" % repr(doc.text))
cdef int arg_max_if_gold(const weight_t* scores, const weight_t* costs, int n) nogil:
cdef int best = -1
for i in range(n):
if costs[i] <= 0:
if best == -1 or scores[i] > scores[best]:
best = i
return best
cdef int _arg_max_clas(const weight_t* scores, int move, const Transition* actions,
int nr_class) except -1:
cdef weight_t score = 0
cdef int mode = -1
cdef int i
for i in range(nr_class):
if actions[i].move == move and (mode == -1 or scores[i] >= score):
mode = i
score = scores[i]
return mode

3
spacy/tests/doc/test_add_entities.py
View File

@ -10,7 +10,8 @@ import pytest
def test_doc_add_entities_set_ents_iob(en_vocab):
text = ["This", "is", "a", "lion"]
doc = get_doc(en_vocab, text)
ner = EntityRecognizer(en_vocab, features=[(2,), (3,)])
ner = EntityRecognizer(en_vocab)
ner.begin_training([])
ner(doc)
assert len(list(doc.ents)) == 0

4
spacy/tests/parser/test_add_label.py
View File

@ -9,7 +9,7 @@ from ...attrs import NORM
from ...gold import GoldParse
from ...vocab import Vocab
from ...tokens import Doc
from ...pipeline import NeuralDependencyParser
from ...pipeline import DependencyParser
numpy.random.seed(0)
@ -21,7 +21,7 @@ def vocab():
@pytest.fixture
def parser(vocab):
parser = NeuralDependencyParser(vocab)
parser = DependencyParser(vocab)
parser.cfg['token_vector_width'] = 8
parser.cfg['hidden_width'] = 30
parser.cfg['hist_size'] = 0

2
spacy/tests/parser/test_neural_parser.py
View File

@ -6,7 +6,7 @@ import numpy
from ..._ml import chain, Tok2Vec, doc2feats
from ...vocab import Vocab
from ...pipeline import TokenVectorEncoder
from ...pipeline import Tensorizer
from ...syntax.arc_eager import ArcEager
from ...syntax.nn_parser import Parser
from ...tokens.doc import Doc

4
spacy/tests/parser/test_preset_sbd.py
View File

@ -8,7 +8,7 @@ from ...attrs import NORM
from ...gold import GoldParse
from ...vocab import Vocab
from ...tokens import Doc
from ...pipeline import NeuralDependencyParser
from ...pipeline import DependencyParser
@pytest.fixture
def vocab():
@ -16,7 +16,7 @@ def vocab():
@pytest.fixture
def parser(vocab):
parser = NeuralDependencyParser(vocab)
parser = DependencyParser(vocab)
parser.cfg['token_vector_width'] = 4
parser.cfg['hidden_width'] = 32
#parser.add_label('right')

6
spacy/tests/parser/test_to_from_bytes_disk.py
View File

@ -1,11 +1,11 @@
import pytest
from ...pipeline import NeuralDependencyParser
from ...pipeline import DependencyParser
@pytest.fixture
def parser(en_vocab):
parser = NeuralDependencyParser(en_vocab)
parser = DependencyParser(en_vocab)
parser.add_label('nsubj')
parser.model, cfg = parser.Model(parser.moves.n_moves)
parser.cfg.update(cfg)
@ -14,7 +14,7 @@ def parser(en_vocab):
@pytest.fixture
def blank_parser(en_vocab):
parser = NeuralDependencyParser(en_vocab)
parser = DependencyParser(en_vocab)
return parser

18
spacy/tests/pipeline/test_pipe_methods.py
View File

@ -82,3 +82,21 @@ def test_remove_pipe(nlp, name):
assert not len(nlp.pipeline)
assert removed_name == name
assert removed_component == new_pipe
@pytest.mark.parametrize('name', ['my_component'])
def test_disable_pipes_method(nlp, name):
nlp.add_pipe(new_pipe, name=name)
assert nlp.has_pipe(name)
disabled = nlp.disable_pipes(name)
assert not nlp.has_pipe(name)
disabled.restore()
@pytest.mark.parametrize('name', ['my_component'])
def test_disable_pipes_context(nlp, name):
nlp.add_pipe(new_pipe, name=name)
assert nlp.has_pipe(name)
with nlp.disable_pipes(name):
assert not nlp.has_pipe(name)
assert nlp.has_pipe(name)

4
spacy/tests/serialize/test_serialize_parser_ner.py
View File

@ -2,8 +2,8 @@
from __future__ import unicode_literals
from ..util import make_tempdir
from ...pipeline import NeuralDependencyParser as DependencyParser
from ...pipeline import NeuralEntityRecognizer as EntityRecognizer
from ...pipeline import DependencyParser
from ...pipeline import EntityRecognizer
import pytest

2
spacy/tests/serialize/test_serialize_tagger.py
View File

@ -2,7 +2,7 @@
from __future__ import unicode_literals
from ..util import make_tempdir
from ...pipeline import NeuralTagger as Tagger
from ...pipeline import Tagger
import pytest

2
spacy/tests/serialize/test_serialize_tensorizer.py
View File

@ -2,7 +2,7 @@
from __future__ import unicode_literals
from ..util import make_tempdir
from ...pipeline import TokenVectorEncoder as Tensorizer
from ...pipeline import Tensorizer
import pytest

1
spacy/vocab.pyx
View File

@ -15,7 +15,6 @@ from .lexeme cimport EMPTY_LEXEME
from .lexeme cimport Lexeme
from .strings cimport hash_string
from .typedefs cimport attr_t
from .cfile cimport CFile
from .tokens.token cimport Token
from .attrs cimport PROB, LANG
from .structs cimport SerializedLexemeC

2
website/_includes/_mixins.jade
View File

@ -181,7 +181,7 @@ mixin codepen(slug, height, default_tab)
alt_file - [string] alternative file path used in footer and link button
height - [integer] height of code preview in px
mixin github(repo, file, alt_file, height, language)
mixin github(repo, file, height, alt_file, language)
- var branch = ALPHA ? "develop" : "master"
- var height = height || 250

2
website/_includes/_page_models.jade
View File

@ -38,7 +38,7 @@ for id in CURRENT_MODELS
+cell #[+label Size]
+cell #[+tag=comps.size] #[span(data-tpl=id data-tpl-key="size") #[em n/a]]
each label in ["Pipeline", "Sources", "Author", "License"]
each label in ["Pipeline", "Vectors", "Sources", "Author", "License"]
- var field = label.toLowerCase()
+row
+cell.u-nowrap

4
website/api/_annotation/_training.jade
View File

@ -13,7 +13,9 @@ p
| that are part of an entity are set to the entity label, prefixed by the
| BILUO marker. For example #[code "B-ORG"] describes the first token of
| a multi-token #[code ORG] entity and #[code "U-PERSON"] a single
| token representing a #[code PERSON] entity
| token representing a #[code PERSON] entity. The
| #[+api("goldparse#biluo_tags_from_offsets") #[code biluo_tags_from_offsets]]
| function can help you convert entity offsets to the right format.
+code("Example structure").
[{

2
website/api/_top-level/_spacy.jade
View File

@ -136,7 +136,7 @@ p
| #[+src(gh("spacy", "spacy/glossary.py")) #[code glossary.py]].
+aside-code("Example").
spacy.explain('NORP')
spacy.explain(u'NORP')
# Nationalities or religious or political groups
doc = nlp(u'Hello world')

1
website/api/dependencyparser.jade
View File

@ -2,4 +2,5 @@
include ../_includes/_mixins
//- This class inherits from Pipe, so this page uses the template in pipe.jade.
!=partial("pipe", { subclass: "DependencyParser", short: "parser", pipeline_id: "parser" })

1
website/api/entityrecognizer.jade
View File

@ -2,4 +2,5 @@
include ../_includes/_mixins
//- This class inherits from Pipe, so this page uses the template in pipe.jade.
!=partial("pipe", { subclass: "EntityRecognizer", short: "ner", pipeline_id: "ner" })

31
website/api/language.jade
View File

@ -441,6 +441,37 @@ p
+cell tuple
+cell A #[code (name, component)] tuple of the removed component.
+h(2, "disable_pipes") Language.disable_pipes
+tag contextmanager
+tag-new(2)
p
| Disable one or more pipeline components. If used as a context manager,
| the pipeline will be restored to the initial state at the end of the
| block. Otherwise, a #[code DisabledPipes] object is returned, that has a
| #[code .restore()] method you can use to undo your changes.
+aside-code("Example").
with nlp.disable_pipes('tagger', 'parser'):
optimizer = nlp.begin_training(gold_tuples)
disabled = nlp.disable_pipes('tagger', 'parser')
optimizer = nlp.begin_training(gold_tuples)
disabled.restore()
+table(["Name", "Type", "Description"])
+row
+cell #[code *disabled]
+cell unicode
+cell Names of pipeline components to disable.
+row("foot")
+cell returns
+cell #[code DisabledPipes]
+cell
| The disabled pipes that can be restored by calling the object's
| #[code .restore()] method.
+h(2, "to_disk") Language.to_disk
+tag method
+tag-new(2)

15
website/api/pipe.jade
View File

@ -304,6 +304,21 @@ p Modify the pipe's model, to use the given parameter values.
| The parameter values to use in the model. At the end of the
| context, the original parameters are restored.
+h(2, "add_label") #{CLASSNAME}.add_label
+tag method
p Add a new label to the pipe.
+aside-code("Example").
#{VARNAME} = #{CLASSNAME}(nlp.vocab)
#{VARNAME}.add_label('MY_LABEL')
+table(["Name", "Type", "Description"])
+row
+cell #[code label]
+cell unicode
+cell The label to add.
+h(2, "to_disk") #{CLASSNAME}.to_disk
+tag method

1
website/api/tagger.jade
View File

@ -2,4 +2,5 @@
include ../_includes/_mixins
//- This class inherits from Pipe, so this page uses the template in pipe.jade.
!=partial("pipe", { subclass: "Tagger", pipeline_id: "tagger" })

1
website/api/tensorizer.jade
View File

@ -2,4 +2,5 @@
include ../_includes/_mixins
//- This class inherits from Pipe, so this page uses the template in pipe.jade.
!=partial("pipe", { subclass: "Tensorizer", pipeline_id: "tensorizer" })

1
website/api/textcategorizer.jade
View File

@ -16,4 +16,5 @@ p
| before a logistic activation is applied elementwise. The value of each
| output neuron is the probability that some class is present.
//- This class inherits from Pipe, so this page uses the template in pipe.jade.
!=partial("pipe", { subclass: "TextCategorizer", short: "textcat", pipeline_id: "textcat" })

11
website/assets/js/main.js
View File

@ -140,6 +140,10 @@ class ModelLoader {
else return ({ ok: res.ok })
}
convertNumber(num, separator = ',') {
return num.toString().replace(/\B(?=(\d{3})+(?!\d))/g, separator);
}
getModels(compat) {
this.compat = compat;
for (let modelId of this.modelIds) {
@ -159,7 +163,7 @@ class ModelLoader {
const template = new Templater(modelId);
template.get('table').removeAttribute('data-loading');
template.get('error').style.display = 'block';
for (let key of ['sources', 'pipeline', 'author', 'license']) {
for (let key of ['sources', 'pipeline', 'vectors', 'author', 'license']) {
template.get(key).parentElement.parentElement.style.display = 'none';
}
}
@ -167,13 +171,14 @@ class ModelLoader {
/**
* Update model details in tables. Currently quite hacky :(
*/
render({ lang, name, version, sources, pipeline, url, author, license, accuracy, size, description, notes }) {
render({ lang, name, version, sources, pipeline, vectors, url, author, license, accuracy, size, description, notes }) {
const modelId = `${lang}_${name}`;
const model = `${modelId}-${version}`;
const template = new Templater(modelId);
const getSources = s => (s instanceof Array) ? s.join(', ') : s;
const getPipeline = p => p.map(comp => `<code>${comp}</code>`).join(', ');
const getVectors = v => `${this.convertNumber(v.entries)} (${v.width} dimensions)`;
const getLink = (t, l) => `<a href="${l}" target="_blank">${t}</a>`;
const keys = { version, size, description, notes }
@ -182,6 +187,8 @@ class ModelLoader {
if (sources) template.fill('sources', getSources(sources));
if (pipeline && pipeline.length) template.fill('pipeline', getPipeline(pipeline), true);
else template.get('pipeline').parentElement.parentElement.style.display = 'none';
if (vectors) template.fill('vectors', getVectors(vectors));
else template.get('vectors').parentElement.parentElement.style.display = 'none';
if (author) template.fill('author', url ? getLink(author, url) : author, true);
if (license) template.fill('license', this.licenses[license] ? getLink(license, this.licenses[license]) : license, true);

5
website/usage/_data.json
View File

@ -106,7 +106,7 @@
"How Pipelines Work": "pipelines",
"Custom Components": "custom-components",
"Developing Extensions": "extensions",
"Multi-threading": "multithreading",
"Multi-Threading": "multithreading",
"Serialization": "serialization"
}
},
@ -196,9 +196,10 @@
"teaser": "Full code examples you can modify and run.",
"next": "resources",
"menu": {
"Information Extraction": "information-extraction",
"Pipeline": "pipeline",
"Matching": "matching",
"Training": "training",
"Vectors & Similarity": "vectors",
"Deep Learning": "deep-learning"
}
}

4
website/usage/_processing-pipelines/_custom-components.jade
View File

@ -234,7 +234,7 @@ p
| when you customise spaCy's tokenization rules. When you call #[code nlp]
| on a text, the custom pipeline component is applied to the #[code Doc]
+github("spacy", "examples/pipeline/custom_component_entities.py", false, 500)
+github("spacy", "examples/pipeline/custom_component_entities.py", 500)
p
| Wrapping this functionality in a
@ -255,7 +255,7 @@ p
| #[code Token] – for example, the capital, latitude/longitude coordinates
| and even the country flag.
+github("spacy", "examples/pipeline/custom_component_countries_api.py", false, 500)
+github("spacy", "examples/pipeline/custom_component_countries_api.py", 500)
p
| In this case, all data can be fetched on initialisation in one request.

13
website/usage/_processing-pipelines/_multithreading.jade
View File

@ -38,3 +38,16 @@ p
| the generator in two, and then #[code izip] the extra stream to the
| document stream. Here's
| #[+a(gh("spacy") + "/issues/172#issuecomment-183963403") an example].
+h(3, "multi-processing-example") Example: Multi-processing with Joblib
p
| This example shows how to use multiple cores to process text using
| spaCy and #[+a("https://pythonhosted.org/joblib/") Joblib]. We're
| exporting part-of-speech-tagged, true-cased, (very roughly)
| sentence-separated text, with each "sentence" on a newline, and
| spaces between tokens. Data is loaded from the IMDB movie reviews
| dataset and will be loaded automatically via Thinc's built-in dataset
| loader.
+github("spacy", "examples/pipeline/multi_processing.py", 500)

116
website/usage/_training/_ner.jade
View File

@ -24,38 +24,108 @@ p
| #[strong experiment on your own data] to find a solution that works best
| for you.
+h(3, "example-new-entity-type") Example: Training an additional entity type
+h(3, "example-train-ner") Updating the Named Entity Recognizer
p
| This script shows how to add a new entity type to an existing pre-trained
| NER model. To keep the example short and simple, only a few sentences are
| This example shows how to update spaCy's entity recognizer
| with your own examples, starting off with an existing, pre-trained
| model, or from scratch using a blank #[code Language] class. To do
| this, you'll need #[strong example texts] and the
| #[strong character offsets] and #[strong labels] of each entity contained
| in the texts.
+github("spacy", "examples/training/train_ner.py", 500)
+h(4) Step by step guide
+list("numbers")
+item
| #[strong Reformat the training data] to match spaCy's
| #[+a("/api/annotation#json-input") JSON format]. The built-in
| #[+api("goldparse#biluo_tags_from_offsets") #[code biluo_tags_from_offsets]]
| function can help you with this.
+item
| #[strong Load the model] you want to start with, or create an
| #[strong empty model] using
| #[+api("spacy#blank") #[code spacy.blank]] with the ID of your
| language. If you're using a blank model, don't forget to add the
| entity recognizer to the pipeline. If you're using an existing model,
| make sure to disable all other pipeline components during training
| using #[+api("language#disable_pipes") #[code nlp.disable_pipes]].
| This way, you'll only be training the entity recognizer.
+item
| #[strong Shuffle and loop over] the examples and create a
| #[code Doc] and #[code GoldParse] object for each example.
+item
| For each example, #[strong update the model]
| by calling #[+api("language#update") #[code nlp.update]], which steps
| through the words of the input. At each word, it makes a
| #[strong prediction]. It then consults the annotations provided on the
| #[code GoldParse] instance, to see whether it was
| right. If it was wrong, it adjusts its weights so that the correct
| action will score higher next time.
+item
| #[strong Save] the trained model using
| #[+api("language#to_disk") #[code nlp.to_disk]].
+item
| #[strong Test] the model to make sure the entities in the training
| data are recognised correctly.
+h(3, "example-new-entity-type") Training an additional entity type
p
| This script shows how to add a new entity type #[code ANIMAL] to an
| existing pre-trained NER model, or an empty #[code Language] class. To
| keep the example short and simple, only a few sentences are
| provided as examples. In practice, you'll need many more — a few hundred
| would be a good start. You will also likely need to mix in examples of
| other entity types, which might be obtained by running the entity
| recognizer over unlabelled sentences, and adding their annotations to the
| training set.
p
| The actual training is performed by looping over the examples, and
| calling #[+api("language#update") #[code nlp.update()]]. The
| #[code update] method steps through the words of the input. At each word,
| it makes a prediction. It then consults the annotations provided on the
| #[+api("goldparse") #[code GoldParse]] instance, to see whether it was
| right. If it was wrong, it adjusts its weights so that the correct
| action will score higher next time.
+github("spacy", "examples/training/train_new_entity_type.py", 500)
+github("spacy", "examples/training/train_new_entity_type.py")
+h(4) Step by step guide
+h(3, "example-ner-from-scratch") Example: Training an NER system from scratch
+list("numbers")
+item
| Create #[code Doc] and #[code GoldParse] objects for
| #[strong each example in your training data].
p
| This example is written to be self-contained and reasonably transparent.
| To achieve that, it duplicates some of spaCy's internal functionality.
| Specifically, in this example, we don't use spaCy's built-in
| #[+api("language") #[code Language]] class to wire together the
| #[+api("vocab") #[code Vocab]], #[+api("tokenizer") #[code Tokenizer]]
| and #[+api("entityrecognizer") #[code EntityRecognizer]]. Instead, we
| write our own simle #[code Pipeline] class, so that it's easier to see
| how the pieces interact.
+item
| #[strong Load the model] you want to start with, or create an
| #[strong empty model] using
| #[+api("spacy#blank") #[code spacy.blank]] with the ID of your
| language. If you're using a blank model, don't forget to add the
| entity recognizer to the pipeline. If you're using an existing model,
| make sure to disable all other pipeline components during training
| using #[+api("language#disable_pipes") #[code nlp.disable_pipes]].
| This way, you'll only be training the entity recognizer.
+github("spacy", "examples/training/train_ner_standalone.py")
+item
| #[strong Add the new entity label] to the entity recognizer using the
| #[+api("entityrecognizer#add_label") #[code add_label]] method. You
| can access the entity recognizer in the pipeline via
| #[code nlp.get_pipe('ner')].
+item
| #[strong Loop over] the examples and call
| #[+api("language#update") #[code nlp.update]], which steps through
| the words of the input. At each word, it makes a
| #[strong prediction]. It then consults the annotations provided on the
| #[code GoldParse] instance, to see whether it was right. If it was
| wrong, it adjusts its weights so that the correct action will score
| higher next time.
+item
| #[strong Save] the trained model using
| #[+api("language#to_disk") #[code nlp.to_disk]].
+item
| #[strong Test] the model to make sure the new entity is recognised
| correctly.

191
website/usage/_training/_tagger-parser.jade
View File

@ -1,6 +1,195 @@
//- 💫 DOCS > USAGE > TRAINING > TAGGER & PARSER
+under-construction
+h(3, "example-train-parser") Updating the Dependency Parser
p
| This example shows how to train spaCy's dependency parser, starting off
| with an existing model or a blank model. You'll need a set of
| #[strong training examples] and the respective #[strong heads] and
| #[strong dependency label] for each token of the example texts.
+github("spacy", "examples/training/train_parser.py", 500)
+h(4) Step by step guide
+list("numbers")
+item
| #[strong Load the model] you want to start with, or create an
| #[strong empty model] using
| #[+api("spacy#blank") #[code spacy.blank]] with the ID of your
| language. If you're using a blank model, don't forget to add the
| parser to the pipeline. If you're using an existing model,
| make sure to disable all other pipeline components during training
| using #[+api("language#disable_pipes") #[code nlp.disable_pipes]].
| This way, you'll only be training the parser.
+item
| #[strong Add the dependency labels] to the parser using the
| #[+api("dependencyparser#add_label") #[code add_label]] method. If
| you're starting off with a pre-trained spaCy model, this is usually
| not necessary – but it doesn't hurt either, just to be safe.
+item
| #[strong Shuffle and loop over] the examples and create a
| #[code Doc] and #[code GoldParse] object for each example. Make sure
| to pass in the #[code heads] and #[code deps] when you create the
| #[code GoldParse].
+item
| For each example, #[strong update the model]
| by calling #[+api("language#update") #[code nlp.update]], which steps
| through the words of the input. At each word, it makes a
| #[strong prediction]. It then consults the annotations provided on the
| #[code GoldParse] instance, to see whether it was
| right. If it was wrong, it adjusts its weights so that the correct
| action will score higher next time.
+item
| #[strong Save] the trained model using
| #[+api("language#to_disk") #[code nlp.to_disk]].
+item
| #[strong Test] the model to make sure the parser works as expected.
+h(3, "example-train-tagger") Updating the Part-of-speech Tagger
p
| In this example, we're training spaCy's part-of-speech tagger with a
| custom tag map. We start off with a blank #[code Language] class, update
| its defaults with our custom tags and then train the tagger. You'll need
| a set of #[strong training examples] and the respective
| #[strong custom tags], as well as a dictionary mapping those tags to the
| #[+a("http://universaldependencies.github.io/docs/u/pos/index.html") Universal Dependencies scheme].
+github("spacy", "examples/training/train_tagger.py", 500)
+h(4) Step by step guide
+list("numbers")
+item
| #[strong Create] a new #[code Language] class and before initialising
| it, update the #[code tag_map] in its #[code Defaults] with your
| custom tags.
+item
| #[strong Create a new tagger] component and add it to the pipeline.
+item
| #[strong Shuffle and loop over] the examples and create a
| #[code Doc] and #[code GoldParse] object for each example. Make sure
| to pass in the #[code tags] when you create the #[code GoldParse].
+item
| For each example, #[strong update the model]
| by calling #[+api("language#update") #[code nlp.update]], which steps
| through the words of the input. At each word, it makes a
| #[strong prediction]. It then consults the annotations provided on the
| #[code GoldParse] instance, to see whether it was
| right. If it was wrong, it adjusts its weights so that the correct
| action will score higher next time.
+item
| #[strong Save] the trained model using
| #[+api("language#to_disk") #[code nlp.to_disk]].
+item
| #[strong Test] the model to make sure the parser works as expected.
+h(3, "intent-parser") Training a parser for custom semantics
p
| spaCy's parser component can be used to trained to predict any type
| of tree structure over your input text – including
| #[strong semantic relations] that are not syntactic dependencies. This
| can be useful to for #[strong conversational applications], which need to
| predict trees over whole documents or chat logs, with connections between
| the sentence roots used to annotate discourse structure. For example, you
| can train spaCy's parser to label intents and their targets, like
| attributes, quality, time and locations. The result could look like this:
+codepen("991f245ef90debb78c8fc369294f75ad", 300)
+code.
doc = nlp(u"find a hotel with good wifi")
print([(t.text, t.dep_, t.head.text) for t in doc if t.dep_ != '-'])
# [('find', 'ROOT', 'find'), ('hotel', 'PLACE', 'find'),
# ('good', 'QUALITY', 'wifi'), ('wifi', 'ATTRIBUTE', 'hotel')]
p
| The above tree attaches "wifi" to "hotel" and assigns the dependency
| label #[code ATTRIBUTE]. This may not be a correct syntactic dependency –
| but in this case, it expresses exactly what we need: the user is looking
| for a hotel with the attribute "wifi" of the quality "good". This query
| can then be processed by your application and used to trigger the
| respective action – e.g. search the database for hotels with high ratings
| for their wifi offerings.
+aside("Tip: merge phrases and entities")
| To achieve even better accuracy, try merging multi-word tokens and
| entities specific to your domain into one token before parsing your text.
| You can do this by running the entity recognizer or
| #[+a("/usage/linguistic-features#rule-based-matching") rule-based matcher]
| to find relevant spans, and merging them using
| #[+api("span#merge") #[code Span.merge]]. You could even add your own
| custom #[+a("/usage/processing-pipelines#custom-components") pipeline component]
| to do this automatically – just make sure to add it #[code before='parser'].
p
| The following example example shows a full implementation of a training
| loop for a custom message parser for a common "chat intent": finding
| local businesses. Our message semantics will have the following types
| of relations: #[code ROOT], #[code PLACE], #[code QUALITY],
| #[code ATTRIBUTE], #[code TIME] and #[code LOCATION].
+github("spacy", "examples/training/train_intent_parser.py", 500)
+h(4) Step by step guide
+list("numbers")
+item
| #[strong Create the training data] consisting of words, their heads
| and their dependency labels in order. A token's head is the index
| of the token it is attached to. The heads don't need to be
| syntactically correct – they should express the
| #[strong semantic relations] you want the parser to learn. For words
| that shouldn't receive a label, you can choose an arbitrary
| placeholder, for example #[code -].
+item
| #[strong Load the model] you want to start with, or create an
| #[strong empty model] using
| #[+api("spacy#blank") #[code spacy.blank]] with the ID of your
| language. If you're using a blank model, don't forget to add the
| parser to the pipeline. If you're using an existing model,
| make sure to disable all other pipeline components during training
| using #[+api("language#disable_pipes") #[code nlp.disable_pipes]].
| This way, you'll only be training the parser.
+item
| #[strong Add the dependency labels] to the parser using the
| #[+api("dependencyparser#add_label") #[code add_label]] method.
+item
| #[strong Shuffle and loop over] the examples and create a
| #[code Doc] and #[code GoldParse] object for each example. Make sure
| to pass in the #[code heads] and #[code deps] when you create the
| #[code GoldParse].
+item
| For each example, #[strong update the model]
| by calling #[+api("language#update") #[code nlp.update]], which steps
| through the words of the input. At each word, it makes a
| #[strong prediction]. It then consults the annotations provided on the
| #[code GoldParse] instance, to see whether it was
| right. If it was wrong, it adjusts its weights so that the correct
| action will score higher next time.
+item
| #[strong Save] the trained model using
| #[+api("language#to_disk") #[code nlp.to_disk]].
+item
| #[strong Test] the model to make sure the parser works as expected.
+h(3, "training-json") JSON format for training

63
website/usage/_training/_textcat.jade
View File

@ -1,13 +1,62 @@
//- 💫 DOCS > USAGE > TRAINING > TEXT CLASSIFICATION
+under-construction
+h(3, "example-textcat") Example: Training spaCy's text classifier
+h(3, "example-textcat") Adding a text classifier to a spaCy model
+tag-new(2)
p
| This example shows how to use and train spaCy's new
| #[+api("textcategorizer") #[code TextCategorizer]] pipeline component
| on IMDB movie reviews.
| This example shows how to train a multi-label convolutional neural
| network text classifier on IMDB movie reviews, using spaCy's new
| #[+api("textcategorizer") #[code TextCategorizer]] component. The
| dataset will be loaded automatically via Thinc's built-in dataset
| loader. Predictions are available via
| #[+api("doc#attributes") #[code Doc.cats]].
+github("spacy", "examples/training/train_textcat.py")
+github("spacy", "examples/training/train_textcat.py", 500)
+h(4) Step by step guide
+list("numbers")
+item
| #[strong Load the model] you want to start with, or create an
| #[strong empty model] using
| #[+api("spacy#blank") #[code spacy.blank]] with the ID of your
| language. If you're using an existing model, make sure to disable all
| other pipeline components during training using
| #[+api("language#disable_pipes") #[code nlp.disable_pipes]]. This
| way, you'll only be training the text classifier.
+item
| #[strong Add the text classifier] to the pipeline, and add the labels
| you want to train – for example, #[code POSITIVE].
+item
| #[strong Load and pre-process the dataset], shuffle the data and
| split off a part of it to hold back for evaluation. This way, you'll
| be able to see results on each training iteration.
+item
| #[strong Loop over] the training examples, partition them into
| batches and create #[code Doc] and #[code GoldParse] objects for each
| example in the batch.
+item
| #[strong Update the model] by calling
| #[+api("language#update") #[code nlp.update]], which steps
| through the examples and makes a #[strong prediction]. It then
| consults the annotations provided on the #[code GoldParse] instance,
| to see whether it was right. If it was wrong, it adjusts its weights
| so that the correct prediction will score higher next time.
+item
| Optionally, you can also #[strong evaluate the text classifier] on
| each iteration, by checking how it performs on the development data
| held back from the dataset. This lets you print the
| #[strong precision], #[strong recall] and #[strong F-score].
+item
| #[strong Save] the trained model using
| #[+api("language#to_disk") #[code nlp.to_disk]].
+item
| #[strong Test] the model to make sure the text classifier works as
| expected.

121
website/usage/examples.jade
View File

@ -2,6 +2,37 @@
include ../_includes/_mixins
+section("information-extraction")
+h(3, "phrase-matcher") Using spaCy's phrase matcher
+tag-new(2)
p
| This example shows how to use the new
| #[+api("phrasematcher") #[code PhraseMatcher]] to efficiently find
| entities from a large terminology list.
+github("spacy", "examples/information_extraction/phrase_matcher.py")
+h(3, "entity-relations") Extracting entity relations
p
| A simple example of extracting relations between phrases and
| entities using spaCy's named entity recognizer and the dependency
| parse. Here, we extract money and currency values (entities labelled
| as #[code MONEY]) and then check the dependency tree to find the
| noun phrase they are referring to – for example: "$9.4 million"
| &rarr; "Net income".
+github("spacy", "examples/information_extraction/entity_relations.py")
+h(3, "subtrees") Navigating the parse tree and subtrees
p
| This example shows how to navigate the parse tree including subtrees
| attached to a word.
+github("spacy", "examples/information_extraction/parse_subtrees.py")
+section("pipeline")
+h(3, "custom-components-entities") Custom pipeline components and attribute extensions
+tag-new(2)
@ -40,27 +71,29 @@ include ../_includes/_mixins
+github("spacy", "examples/pipeline/custom_attr_methods.py")
+section("matching")
+h(3, "matcher") Using spaCy's rule-based matcher
+h(3, "multi-processing") Multi-processing with Joblib
p
| This example shows how to use spaCy's rule-based
| #[+api("matcher") #[code Matcher]] to find and label entities across
| documents.
| This example shows how to use multiple cores to process text using
| spaCy and #[+a("https://pythonhosted.org/joblib/") Joblib]. We're
| exporting part-of-speech-tagged, true-cased, (very roughly)
| sentence-separated text, with each "sentence" on a newline, and
| spaces between tokens. Data is loaded from the IMDB movie reviews
| dataset and will be loaded automatically via Thinc's built-in dataset
| loader.
+github("spacy", "examples/matcher_example.py")
+h(3, "phrase-matcher") Using spaCy's phrase matcher
+tag-new(2)
p
| This example shows how to use the new
| #[+api("phrasematcher") #[code PhraseMatcher]] to efficiently find
| entities from a large terminology list.
+github("spacy", "examples/phrase_matcher.py")
+github("spacy", "examples/pipeline/multi_processing.py")
+section("training")
+h(3, "training-ner") Training spaCy's Named Entity Recognizer
p
| This example shows how to update spaCy's entity recognizer
| with your own examples, starting off with an existing, pre-trained
| model, or from scratch using a blank #[code Language] class.
+github("spacy", "examples/training/train_ner.py")
+h(3, "new-entity-type") Training an additional entity type
p
@ -71,25 +104,63 @@ include ../_includes/_mixins
+github("spacy", "examples/training/train_new_entity_type.py")
+h(3, "ner-standalone") Training an NER system from scratch
+h(3, "parser") Training spaCy's Dependency Parser
p
| This example is written to be self-contained and reasonably
| transparent. To achieve that, it duplicates some of spaCy's internal
| functionality.
| This example shows how to update spaCy's dependency parser,
| starting off with an existing, pre-trained model, or from scratch
| using a blank #[code Language] class.
+github("spacy", "examples/training/train_ner_standalone.py")
+github("spacy", "examples/training/train_parser.py")
+h(3, "tagger") Training spaCy's Part-of-speech Tagger
p
| In this example, we're training spaCy's part-of-speech tagger with a
| custom tag map, mapping our own tags to the mapping those tags to the
| #[+a("http://universaldependencies.github.io/docs/u/pos/index.html") Universal Dependencies scheme].
+github("spacy", "examples/training/train_tagger.py")
+h(3, "intent-parser") Training a custom parser for chat intent semantics
p
| spaCy's parser component can be used to trained to predict any type
| of tree structure over your input text. You can also predict trees
| over whole documents or chat logs, with connections between the
| sentence-roots used to annotate discourse structure. In this example,
| we'll build a message parser for a common "chat intent": finding
| local businesses. Our message semantics will have the following types
| of relations: #[code ROOT], #[code PLACE], #[code QUALITY],
| #[code ATTRIBUTE], #[code TIME] and #[code LOCATION].
+github("spacy", "examples/training/train_intent_parser.py")
+h(3, "textcat") Training spaCy's text classifier
+tag-new(2)
p
| This example shows how to use and train spaCy's new
| #[+api("textcategorizer") #[code TextCategorizer]] pipeline component
| on IMDB movie reviews.
| This example shows how to train a multi-label convolutional neural
| network text classifier on IMDB movie reviews, using spaCy's new
| #[+api("textcategorizer") #[code TextCategorizer]] component. The
| dataset will be loaded automatically via Thinc's built-in dataset
| loader. Predictions are available via
| #[+api("doc#attributes") #[code Doc.cats]].
+github("spacy", "examples/training/train_textcat.py")
+section("vectors")
+h(3, "fasttext") Loading pre-trained fastText vectors
p
| This simple snippet is all you need to be able to use the Facebook's
| #[+a("https://github.com/facebookresearch/fastText/blob/master/pretrained-vectors.md") fastText vectors]
| (294 languages, pre-trained on Wikipedia) with spaCy. Once they're
| loaded, the vectors will be available via spaCy's built-in
| #[code similarity()] methods.
+github("spacy", "examples/vectors_fast_text.py")
+section("deep-learning")
+h(3, "keras") Text classification with Keras
@ -108,4 +179,4 @@ include ../_includes/_mixins
| parameters, and was implemented using #[+a("https://keras.io") Keras]
| and spaCy.
+github("spacy", "examples/keras_parikh_entailment/__main__.py", "examples/keras_parikh_entailment")
+github("spacy", "examples/keras_parikh_entailment/__main__.py", false, "examples/keras_parikh_entailment")

6
website/usage/text-classification.jade
View File

@ -2,8 +2,4 @@
include ../_includes/_mixins
+under-construction
+h(2, "example") Example
+github("spacy", "examples/training/train_textcat.py")
include _training/_textcat