explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-12-24 17:06:29 +03:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into spacy.io

Browse Source
This commit is contained in:
Ines Montani 2019-03-25 13:05:00 +01:00
commit 21ade53ef7
24 changed files with 511 additions and 260 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
examples/training/train_textcat.py
View File

@ -43,7 +43,11 @@ def main(model=None, output_dir=None, n_iter=20, n_texts=2000, init_tok2vec=None
# nlp.create_pipe works for built-ins that are registered with spaCy
if "textcat" not in nlp.pipe_names:
textcat = nlp.create_pipe(
"textcat", config={"architecture": "simple_cnn", "exclusive_classes": True}
"textcat",
config={
"exclusive_classes": True,
"architecture": "simple_cnn",
}
)
nlp.add_pipe(textcat, last=True)
# otherwise, get it, so we can add labels to it
@ -56,7 +60,9 @@ def main(model=None, output_dir=None, n_iter=20, n_texts=2000, init_tok2vec=None
# load the IMDB dataset
print("Loading IMDB data...")
(train_texts, train_cats), (dev_texts, dev_cats) = load_data(limit=n_texts)
(train_texts, train_cats), (dev_texts, dev_cats) = load_data()
train_texts = train_texts[:n_texts]
train_cats = train_cats[:n_texts]
print(
"Using {} examples ({} training, {} evaluation)".format(
n_texts, len(train_texts), len(dev_texts)

5
netlify.toml
View File

@ -43,8 +43,9 @@ redirects = [
{from = "/usage/lightning-tour", to = "/usage/spacy-101#lightning-tour"},
{from = "/usage/linguistic-features#rule-based-matching", to = "/usage/rule-based-matching"},
{from = "/models/comparison", to = "/models"},
{from = "/api/#section-cython", to = "/api/cython"},
{from = "/api/#cython", to = "/api/cython"},
{from = "/api/#section-cython", to = "/api/cython", force = true},
{from = "/api/#cython", to = "/api/cython", force = true},
{from = "/api/sentencesegmenter", to="/api/sentencizer"},
{from = "/universe", to = "/universe/project/:id", query = {id = ":id"}, force = true},
{from = "/universe", to = "/universe/category/:category", query = {category = ":category"}, force = true},
]

94
spacy/_ml.py
View File

@ -81,18 +81,6 @@ def _zero_init(model):
return model
@layerize
def _preprocess_doc(docs, drop=0.0):
keys = [doc.to_array(LOWER) for doc in docs]
# The dtype here matches what thinc is expecting -- which differs per
# platform (by int definition). This should be fixed once the problem
# is fixed on Thinc's side.
lengths = numpy.array([arr.shape[0] for arr in keys], dtype=numpy.int_)
keys = numpy.concatenate(keys)
vals = numpy.zeros(keys.shape, dtype='f')
return (keys, vals, lengths), None
def with_cpu(ops, model):
"""Wrap a model that should run on CPU, transferring inputs and outputs
as necessary."""
@ -133,20 +121,31 @@ def _to_device(ops, X):
return ops.asarray(X)
@layerize
def _preprocess_doc_bigrams(docs, drop=0.0):
unigrams = [doc.to_array(LOWER) for doc in docs]
ops = Model.ops
bigrams = [ops.ngrams(2, doc_unis) for doc_unis in unigrams]
keys = [ops.xp.concatenate(feats) for feats in zip(unigrams, bigrams)]
keys, vals = zip(*[ops.xp.unique(k, return_counts=True) for k in keys])
# The dtype here matches what thinc is expecting -- which differs per
# platform (by int definition). This should be fixed once the problem
# is fixed on Thinc's side.
lengths = ops.asarray([arr.shape[0] for arr in keys], dtype=numpy.int_)
keys = ops.xp.concatenate(keys)
vals = ops.asarray(ops.xp.concatenate(vals), dtype="f")
return (keys, vals, lengths), None
class extract_ngrams(Model):
def __init__(self, ngram_size, attr=LOWER):
Model.__init__(self)
self.ngram_size = ngram_size
self.attr = attr
def begin_update(self, docs, drop=0.0):
batch_keys = []
batch_vals = []
for doc in docs:
unigrams = doc.to_array([self.attr])
ngrams = [unigrams]
for n in range(2, self.ngram_size + 1):
ngrams.append(self.ops.ngrams(n, unigrams))
keys = self.ops.xp.concatenate(ngrams)
keys, vals = self.ops.xp.unique(keys, return_counts=True)
batch_keys.append(keys)
batch_vals.append(vals)
# The dtype here matches what thinc is expecting -- which differs per
# platform (by int definition). This should be fixed once the problem
# is fixed on Thinc's side.
lengths = self.ops.asarray([arr.shape[0] for arr in batch_keys], dtype=numpy.int_)
batch_keys = self.ops.xp.concatenate(batch_keys)
batch_vals = self.ops.asarray(self.ops.xp.concatenate(batch_vals), dtype="f")
return (batch_keys, batch_vals, lengths), None
@describe.on_data(
@ -486,16 +485,6 @@ def zero_init(model):
return model
@layerize
def preprocess_doc(docs, drop=0.0):
keys = [doc.to_array([LOWER]) for doc in docs]
ops = Model.ops
lengths = ops.asarray([arr.shape[0] for arr in keys])
keys = ops.xp.concatenate(keys)
vals = ops.allocate(keys.shape[0]) + 1
return (keys, vals, lengths), None
def getitem(i):
def getitem_fwd(X, drop=0.0):
return X[i], None
@ -602,10 +591,8 @@ def build_text_classifier(nr_class, width=64, **cfg):
>> zero_init(Affine(nr_class, width, drop_factor=0.0))
)
linear_model = (
_preprocess_doc
>> with_cpu(Model.ops, LinearModel(nr_class))
)
linear_model = build_bow_text_classifier(
nr_class, ngram_size=cfg.get("ngram_size", 1), exclusive_classes=False)
if cfg.get('exclusive_classes'):
output_layer = Softmax(nr_class, nr_class * 2)
else:
@ -623,6 +610,33 @@ def build_text_classifier(nr_class, width=64, **cfg):
return model
def build_bow_text_classifier(nr_class, ngram_size=1, exclusive_classes=False,
no_output_layer=False, **cfg):
with Model.define_operators({">>": chain}):
model = (
extract_ngrams(ngram_size, attr=ORTH)
>> with_cpu(Model.ops,
LinearModel(nr_class)
)
)
if not no_output_layer:
model = model >> (cpu_softmax if exclusive_classes else logistic)
model.nO = nr_class
return model
@layerize
def cpu_softmax(X, drop=0.):
ops = NumpyOps()
Y = ops.softmax(X)
def cpu_softmax_backward(dY, sgd=None):
return dY
return ops.softmax(X), cpu_softmax_backward
def build_simple_cnn_text_classifier(tok2vec, nr_class, exclusive_classes=False, **cfg):
"""
Build a simple CNN text classifier, given a token-to-vector model as inputs.

2
spacy/about.py
View File

@ -4,7 +4,7 @@
# fmt: off
__title__ = "spacy"
__version__ = "2.1.2"
__version__ = "2.1.3"
__summary__ = "Industrial-strength Natural Language Processing (NLP) with Python and Cython"
__uri__ = "https://spacy.io"
__author__ = "Explosion AI"

4
spacy/language.py
View File

@ -15,7 +15,7 @@ from .tokenizer import Tokenizer
from .vocab import Vocab
from .lemmatizer import Lemmatizer
from .pipeline import DependencyParser, Tensorizer, Tagger, EntityRecognizer
from .pipeline import SimilarityHook, TextCategorizer, SentenceSegmenter
from .pipeline import SimilarityHook, TextCategorizer, Sentencizer
from .pipeline import merge_noun_chunks, merge_entities, merge_subtokens
from .pipeline import EntityRuler
from .compat import izip, basestring_
@ -119,7 +119,7 @@ class Language(object):
"ner": lambda nlp, **cfg: EntityRecognizer(nlp.vocab, **cfg),
"similarity": lambda nlp, **cfg: SimilarityHook(nlp.vocab, **cfg),
"textcat": lambda nlp, **cfg: TextCategorizer(nlp.vocab, **cfg),
"sentencizer": lambda nlp, **cfg: SentenceSegmenter(nlp.vocab, **cfg),
"sentencizer": lambda nlp, **cfg: Sentencizer(**cfg),
"merge_noun_chunks": lambda nlp, **cfg: merge_noun_chunks,
"merge_entities": lambda nlp, **cfg: merge_entities,
"merge_subtokens": lambda nlp, **cfg: merge_subtokens,

3
spacy/pipeline/__init__.py
View File

@ -2,7 +2,7 @@
from __future__ import unicode_literals
from .pipes import Tagger, DependencyParser, EntityRecognizer
from .pipes import TextCategorizer, Tensorizer, Pipe
from .pipes import TextCategorizer, Tensorizer, Pipe, Sentencizer
from .entityruler import EntityRuler
from .hooks import SentenceSegmenter, SimilarityHook
from .functions import merge_entities, merge_noun_chunks, merge_subtokens
@ -15,6 +15,7 @@ __all__ = [
"Tensorizer",
"Pipe",
"EntityRuler",
"Sentencizer",
"SentenceSegmenter",
"SimilarityHook",
"merge_entities",

2
spacy/pipeline/entityruler.py
View File

@ -191,7 +191,7 @@ class EntityRuler(object):
**kwargs: Other config paramters, mostly for consistency.
RETURNS (EntityRuler): The loaded entity ruler.
DOCS: https://spacy.io/api/entityruler
DOCS: https://spacy.io/api/entityruler#to_disk
"""
path = ensure_path(path)
path = path.with_suffix(".jsonl")

12
spacy/pipeline/hooks.py
View File

@ -15,8 +15,6 @@ class SentenceSegmenter(object):
initialization, or assign a new strategy to the .strategy attribute.
Sentence detection strategies should be generators that take `Doc` objects
and yield `Span` objects for each sentence.
DOCS: https://spacy.io/api/sentencesegmenter
"""
name = "sentencizer"
@ -35,12 +33,12 @@ class SentenceSegmenter(object):
def split_on_punct(doc):
start = 0
seen_period = False
for i, word in enumerate(doc):
if seen_period and not word.is_punct:
yield doc[start : word.i]
start = word.i
for i, token in enumerate(doc):
if seen_period and not token.is_punct:
yield doc[start : token.i]
start = token.i
seen_period = False
elif word.text in [".", "!", "?"]:
elif token.text in [".", "!", "?"]:
seen_period = True
if start < len(doc):
yield doc[start : len(doc)]

91
spacy/pipeline/pipes.pyx
View File

@ -25,6 +25,7 @@ from ..attrs import POS, ID
from ..parts_of_speech import X
from .._ml import Tok2Vec, build_tagger_model
from .._ml import build_text_classifier, build_simple_cnn_text_classifier
from .._ml import build_bow_text_classifier
from .._ml import link_vectors_to_models, zero_init, flatten
from .._ml import masked_language_model, create_default_optimizer
from ..errors import Errors, TempErrors
@ -876,6 +877,8 @@ class TextCategorizer(Pipe):
if cfg.get("architecture") == "simple_cnn":
tok2vec = Tok2Vec(token_vector_width, embed_size, **cfg)
return build_simple_cnn_text_classifier(tok2vec, nr_class, **cfg)
elif cfg.get("architecture") == "bow":
return build_bow_text_classifier(nr_class, **cfg)
else:
return build_text_classifier(nr_class, **cfg)
@ -1058,4 +1061,90 @@ cdef class EntityRecognizer(Parser):
if move[0] in ("B", "I", "L", "U")))
__all__ = ["Tagger", "DependencyParser", "EntityRecognizer", "Tensorizer", "TextCategorizer"]
class Sentencizer(object):
"""Segment the Doc into sentences using a rule-based strategy.
DOCS: https://spacy.io/api/sentencizer
"""
name = "sentencizer"
default_punct_chars = [".", "!", "?"]
def __init__(self, punct_chars=None, **kwargs):
"""Initialize the sentencizer.
punct_chars (list): Punctuation characters to split on. Will be
serialized with the nlp object.
RETURNS (Sentencizer): The sentencizer component.
DOCS: https://spacy.io/api/sentencizer#init
"""
self.punct_chars = punct_chars or self.default_punct_chars
def __call__(self, doc):
"""Apply the sentencizer to a Doc and set Token.is_sent_start.
doc (Doc): The document to process.
RETURNS (Doc): The processed Doc.
DOCS: https://spacy.io/api/sentencizer#call
"""
start = 0
seen_period = False
for i, token in enumerate(doc):
is_in_punct_chars = token.text in self.punct_chars
token.is_sent_start = i == 0
if seen_period and not token.is_punct and not is_in_punct_chars:
doc[start].is_sent_start = True
start = token.i
seen_period = False
elif is_in_punct_chars:
seen_period = True
if start < len(doc):
doc[start].is_sent_start = True
return doc
def to_bytes(self, **kwargs):
"""Serialize the sentencizer to a bytestring.
RETURNS (bytes): The serialized object.
DOCS: https://spacy.io/api/sentencizer#to_bytes
"""
return srsly.msgpack_dumps({"punct_chars": self.punct_chars})
def from_bytes(self, bytes_data, **kwargs):
"""Load the sentencizer from a bytestring.
bytes_data (bytes): The data to load.
returns (Sentencizer): The loaded object.
DOCS: https://spacy.io/api/sentencizer#from_bytes
"""
cfg = srsly.msgpack_loads(bytes_data)
self.punct_chars = cfg.get("punct_chars", self.default_punct_chars)
return self
def to_disk(self, path, exclude=tuple(), **kwargs):
"""Serialize the sentencizer to disk.
DOCS: https://spacy.io/api/sentencizer#to_disk
"""
path = util.ensure_path(path)
path = path.with_suffix(".json")
srsly.write_json(path, {"punct_chars": self.punct_chars})
def from_disk(self, path, exclude=tuple(), **kwargs):
"""Load the sentencizer from disk.
DOCS: https://spacy.io/api/sentencizer#from_disk
"""
path = util.ensure_path(path)
path = path.with_suffix(".json")
cfg = srsly.read_json(path)
self.punct_chars = cfg.get("punct_chars", self.default_punct_chars)
return self
__all__ = ["Tagger", "DependencyParser", "EntityRecognizer", "Tensorizer", "TextCategorizer", "Sentencizer"]

9
spacy/syntax/nn_parser.pyx
View File

@ -574,11 +574,12 @@ cdef class Parser:
cfg.setdefault('min_action_freq', 30)
actions = self.moves.get_actions(gold_parses=get_gold_tuples(),
min_freq=cfg.get('min_action_freq', 30))
previous_labels = dict(self.moves.labels)
for action, labels in self.moves.labels.items():
actions.setdefault(action, {})
for label, freq in labels.items():
if label not in actions[action]:
actions[action][label] = freq
self.moves.initialize_actions(actions)
for action, label_freqs in previous_labels.items():
for label in label_freqs:
self.moves.add_action(action, label)
cfg.setdefault('token_vector_width', 96)
if self.model is True:
self.model, cfg = self.Model(self.moves.n_moves, **cfg)

59
spacy/tests/parser/test_add_label.py
View File

@ -8,7 +8,8 @@ from spacy.attrs import NORM
from spacy.gold import GoldParse
from spacy.vocab import Vocab
from spacy.tokens import Doc
from spacy.pipeline import DependencyParser
from spacy.pipeline import DependencyParser, EntityRecognizer
from spacy.util import fix_random_seed
@pytest.fixture
@ -19,18 +20,6 @@ def vocab():
@pytest.fixture
def parser(vocab):
parser = DependencyParser(vocab)
parser.cfg["token_vector_width"] = 8
parser.cfg["hidden_width"] = 30
parser.cfg["hist_size"] = 0
parser.add_label("left")
parser.begin_training([], **parser.cfg)
sgd = Adam(NumpyOps(), 0.001)
for i in range(10):
losses = {}
doc = Doc(vocab, words=["a", "b", "c", "d"])
gold = GoldParse(doc, heads=[1, 1, 3, 3], deps=["left", "ROOT", "left", "ROOT"])
parser.update([doc], [gold], sgd=sgd, losses=losses)
return parser
@ -38,25 +27,23 @@ def test_init_parser(parser):
pass
# TODO: This is flakey, because it depends on what the parser first learns.
# TODO: This now seems to be implicated in segfaults. Not sure what's up!
@pytest.mark.skip
def _train_parser(parser):
fix_random_seed(1)
parser.add_label("left")
parser.begin_training([], **parser.cfg)
sgd = Adam(NumpyOps(), 0.001)
for i in range(5):
losses = {}
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
gold = GoldParse(doc, heads=[1, 1, 3, 3], deps=["left", "ROOT", "left", "ROOT"])
parser.update([doc], [gold], sgd=sgd, losses=losses)
return parser
def test_add_label(parser):
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
doc = parser(doc)
assert doc[0].head.i == 1
assert doc[0].dep_ == "left"
assert doc[1].head.i == 1
assert doc[2].head.i == 3
assert doc[2].head.i == 3
parser = _train_parser(parser)
parser.add_label("right")
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
doc = parser(doc)
assert doc[0].head.i == 1
assert doc[0].dep_ == "left"
assert doc[1].head.i == 1
assert doc[2].head.i == 3
assert doc[2].head.i == 3
sgd = Adam(NumpyOps(), 0.001)
for i in range(10):
losses = {}
@ -69,3 +56,15 @@ def test_add_label(parser):
doc = parser(doc)
assert doc[0].dep_ == "right"
assert doc[2].dep_ == "left"
def test_add_label_deserializes_correctly():
ner1 = EntityRecognizer(Vocab())
ner1.add_label("C")
ner1.add_label("B")
ner1.add_label("A")
ner1.begin_training([])
ner2 = EntityRecognizer(Vocab()).from_bytes(ner1.to_bytes())
assert ner1.moves.n_moves == ner2.moves.n_moves
for i in range(ner1.moves.n_moves):
assert ner1.moves.get_class_name(i) == ner2.moves.get_class_name(i)

87
spacy/tests/pipeline/test_sentencizer.py Normal file
View File

@ -0,0 +1,87 @@
# coding: utf8
from __future__ import unicode_literals
import pytest
from spacy.pipeline import Sentencizer
from spacy.tokens import Doc
def test_sentencizer(en_vocab):
doc = Doc(en_vocab, words=["Hello", "!", "This", "is", "a", "test", "."])
sentencizer = Sentencizer()
doc = sentencizer(doc)
assert doc.is_sentenced
sent_starts = [t.is_sent_start for t in doc]
assert sent_starts == [True, False, True, False, False, False, False]
assert len(list(doc.sents)) == 2
@pytest.mark.parametrize(
"words,sent_starts,n_sents",
[
# The expected result here is that the duplicate punctuation gets merged
# onto the same sentence and no one-token sentence is created for them.
(
["Hello", "!", ".", "Test", ".", ".", "ok"],
[True, False, False, True, False, False, True],
3,
),
# We also want to make sure ¡ and ¿ aren't treated as sentence end
# markers, even though they're punctuation
(
["¡", "Buen", "día", "!", "Hola", ",", "¿", "qué", "tal", "?"],
[True, False, False, False, True, False, False, False, False, False],
2,
),
# The Token.is_punct check ensures that quotes are handled as well
(
['"', "Nice", "!", '"', "I", "am", "happy", "."],
[True, False, False, False, True, False, False, False],
2,
),
],
)
def test_sentencizer_complex(en_vocab, words, sent_starts, n_sents):
doc = Doc(en_vocab, words=words)
sentencizer = Sentencizer()
doc = sentencizer(doc)
assert doc.is_sentenced
assert [t.is_sent_start for t in doc] == sent_starts
assert len(list(doc.sents)) == n_sents
@pytest.mark.parametrize(
"punct_chars,words,sent_starts,n_sents",
[
(
["~", "?"],
["Hello", "world", "~", "A", ".", "B", "."],
[True, False, False, True, False, False, False],
2,
),
# Even thought it's not common, the punct_chars should be able to
# handle any tokens
(
[".", "ö"],
["Hello", ".", "Test", "ö", "Ok", "."],
[True, False, True, False, True, False],
3,
),
],
)
def test_sentencizer_custom_punct(en_vocab, punct_chars, words, sent_starts, n_sents):
doc = Doc(en_vocab, words=words)
sentencizer = Sentencizer(punct_chars=punct_chars)
doc = sentencizer(doc)
assert doc.is_sentenced
assert [t.is_sent_start for t in doc] == sent_starts
assert len(list(doc.sents)) == n_sents
def test_sentencizer_serialize_bytes(en_vocab):
punct_chars = [".", "~", "+"]
sentencizer = Sentencizer(punct_chars=punct_chars)
assert sentencizer.punct_chars == punct_chars
bytes_data = sentencizer.to_bytes()
new_sentencizer = Sentencizer().from_bytes(bytes_data)
assert new_sentencizer.punct_chars == punct_chars

22
spacy/tests/regression/test_issue3468.py Normal file
View File

@ -0,0 +1,22 @@
# coding: utf8
from __future__ import unicode_literals
import pytest
from spacy.lang.en import English
from spacy.tokens import Doc
def test_issue3468():
"""Test that sentence boundaries are set correctly so Doc.is_sentenced can
be restored after serialization."""
nlp = English()
nlp.add_pipe(nlp.create_pipe("sentencizer"))
doc = nlp("Hello world")
assert doc[0].is_sent_start
assert doc.is_sentenced
assert len(list(doc.sents)) == 1
doc_bytes = doc.to_bytes()
new_doc = Doc(nlp.vocab).from_bytes(doc_bytes)
assert new_doc[0].is_sent_start
assert new_doc.is_sentenced
assert len(list(new_doc.sents)) == 1

2
spacy/tokens/doc.pyx
View File

@ -230,7 +230,7 @@ cdef class Doc:
defined as having at least one of the following:
a) An entry "sents" in doc.user_hooks";
b) sent.is_parsed is set to True;
b) Doc.is_parsed is set to True;
c) At least one token other than the first where sent_start is not None.
"""
if "sents" in self.user_hooks:

1
spacy/tokens/token.pyx
View File

@ -441,6 +441,7 @@ cdef class Token:
property sent_start:
def __get__(self):
"""Deprecated: use Token.is_sent_start instead."""
# Raising a deprecation warning here causes errors for autocomplete
# Handle broken backwards compatibility case: doc[0].sent_start
# was False.

78
website/docs/api/sentencesegmenter.md
View File

@ -1,78 +0,0 @@
---
title: SentenceSegmenter
tag: class
source: spacy/pipeline/hooks.py
---
A simple spaCy hook, to allow custom sentence boundary detection logic that
doesn't require the dependency parse. By default, sentence segmentation is
performed by the [`DependencyParser`](/api/dependencyparser), so the
`SentenceSegmenter` lets you implement a simpler, rule-based strategy that
doesn't require a statistical model to be loaded. The component is also
available via the string name `"sentencizer"`. After initialization, it is
typically added to the processing pipeline using
[`nlp.add_pipe`](/api/language#add_pipe).
## SentenceSegmenter.\_\_init\_\_ {#init tag="method"}
Initialize the sentence segmenter. To change the sentence boundary detection
strategy, pass a generator function `strategy` on initialization, or assign a
new strategy to the `.strategy` attribute. Sentence detection strategies should
be generators that take `Doc` objects and yield `Span` objects for each
sentence.
> #### Example
>
> ```python
> # Construction via create_pipe
> sentencizer = nlp.create_pipe("sentencizer")
>
> # Construction from class
> from spacy.pipeline import SentenceSegmenter
> sentencizer = SentenceSegmenter(nlp.vocab)
> ```
| Name | Type | Description |
| ----------- | ------------------- | ----------------------------------------------------------- |
| `vocab` | `Vocab` | The shared vocabulary. |
| `strategy` | unicode / callable | The segmentation strategy to use. Defaults to `"on_punct"`. |
| **RETURNS** | `SentenceSegmenter` | The newly constructed object. |
## SentenceSegmenter.\_\_call\_\_ {#call tag="method"}
Apply the sentence segmenter on a `Doc`. Typically, this happens automatically
after the component has been added to the pipeline using
[`nlp.add_pipe`](/api/language#add_pipe).
> #### Example
>
> ```python
> from spacy.lang.en import English
>
> nlp = English()
> sentencizer = nlp.create_pipe("sentencizer")
> nlp.add_pipe(sentencizer)
> doc = nlp(u"This is a sentence. This is another sentence.")
> assert list(doc.sents) == 2
> ```
| Name | Type | Description |
| ----------- | ----- | ------------------------------------------------------------ |
| `doc` | `Doc` | The `Doc` object to process, e.g. the `Doc` in the pipeline. |
| **RETURNS** | `Doc` | The modified `Doc` with added sentence boundaries. |
## SentenceSegmenter.split_on_punct {#split_on_punct tag="staticmethod"}
Split the `Doc` on punctuation characters `.`, `!` and `?`. This is the default
strategy used by the `SentenceSegmenter.`
| Name | Type | Description |
| ---------- | ------ | ------------------------------ |
| `doc` | `Doc` | The `Doc` object to process. |
| **YIELDS** | `Span` | The sentences in the document. |
## Attributes {#attributes}
| Name | Type | Description |
| ---------- | -------- | ------------------------------------------------------------------- |
| `strategy` | callable | The segmentation strategy. Can be overwritten after initialization. |

136
website/docs/api/sentencizer.md Normal file
View File

@ -0,0 +1,136 @@
---
title: Sentencizer
tag: class
source: spacy/pipeline/pipes.pyx
---
A simple pipeline component, to allow custom sentence boundary detection logic
that doesn't require the dependency parse. By default, sentence segmentation is
performed by the [`DependencyParser`](/api/dependencyparser), so the
`Sentencizer` lets you implement a simpler, rule-based strategy that doesn't
require a statistical model to be loaded. The component is also available via
the string name `"sentencizer"`. After initialization, it is typically added to
the processing pipeline using [`nlp.add_pipe`](/api/language#add_pipe).
<Infobox title="Important note" variant="warning">
Compared to the previous `SentenceSegmenter` class, the `Sentencizer` component
doesn't add a hook to `doc.user_hooks["sents"]`. Instead, it iterates over the
tokens in the `Doc` and sets the `Token.is_sent_start` property. The
`SentenceSegmenter` is still available if you import it directly:
```python
from spacy.pipeline import SentenceSegmenter
```
</Infobox>
## Sentencizer.\_\_init\_\_ {#init tag="method"}
Initialize the sentencizer.
> #### Example
>
> ```python
> # Construction via create_pipe
> sentencizer = nlp.create_pipe("sentencizer")
>
> # Construction from class
> from spacy.pipeline import Sentencizer
> sentencizer = Sentencizer()
> ```
| Name | Type | Description |
| ------------- | ------------- | ------------------------------------------------------------------------------------------------------ |
| `punct_chars` | list | Optional custom list of punctuation characters that mark sentence ends. Defaults to `[".", "!", "?"].` |
| **RETURNS** | `Sentencizer` | The newly constructed object. |
## Sentencizer.\_\_call\_\_ {#call tag="method"}
Apply the sentencizer on a `Doc`. Typically, this happens automatically after
the component has been added to the pipeline using
[`nlp.add_pipe`](/api/language#add_pipe).
> #### Example
>
> ```python
> from spacy.lang.en import English
>
> nlp = English()
> sentencizer = nlp.create_pipe("sentencizer")
> nlp.add_pipe(sentencizer)
> doc = nlp(u"This is a sentence. This is another sentence.")
> assert list(doc.sents) == 2
> ```
| Name | Type | Description |
| ----------- | ----- | ------------------------------------------------------------ |
| `doc` | `Doc` | The `Doc` object to process, e.g. the `Doc` in the pipeline. |
| **RETURNS** | `Doc` | The modified `Doc` with added sentence boundaries. |
## Sentencizer.to_disk {#to_disk tag="method"}
Save the sentencizer settings (punctuation characters) a directory. Will create
a file `sentencizer.json`. This also happens automatically when you save an
`nlp` object with a sentencizer added to its pipeline.
> #### Example
>
> ```python
> sentencizer = Sentencizer(punct_chars=[".", "?", "!", "。"])
> sentencizer.to_disk("/path/to/sentencizer.jsonl")
> ```
| Name | Type | Description |
| ------ | ---------------- | ---------------------------------------------------------------------------------------------------------------- |
| `path` | unicode / `Path` | A path to a file, which will be created if it doesn't exist. Paths may be either strings or `Path`-like objects. |
## Sentencizer.from_disk {#from_disk tag="method"}
Load the sentencizer settings from a file. Expects a JSON file. This also
happens automatically when you load an `nlp` object or model with a sentencizer
added to its pipeline.
> #### Example
>
> ```python
> sentencizer = Sentencizer()
> sentencizer.from_disk("/path/to/sentencizer.json")
> ```
| Name | Type | Description |
| ----------- | ---------------- | -------------------------------------------------------------------------- |
| `path` | unicode / `Path` | A path to a JSON file. Paths may be either strings or `Path`-like objects. |
| **RETURNS** | `Sentencizer` | The modified `Sentencizer` object. |
## Sentencizer.to_bytes {#to_bytes tag="method"}
Serialize the sentencizer settings to a bytestring.
> #### Example
>
> ```python
> sentencizer = Sentencizer(punct_chars=[".", "?", "!", "。"])
> sentencizer_bytes = sentencizer.to_bytes()
> ```
| Name | Type | Description |
| ----------- | ----- | -------------------- |
| **RETURNS** | bytes | The serialized data. |
## Sentencizer.from_bytes {#from_bytes tag="method"}
Load the pipe from a bytestring. Modifies the object in place and returns it.
> #### Example
>
> ```python
> sentencizer_bytes = sentencizer.to_bytes()
> sentencizer = Sentencizer()
> sentencizer.from_bytes(sentencizer_bytes)
> ```
| Name | Type | Description |
| ------------ | ------------- | ---------------------------------- |
| `bytes_data` | bytes | The bytestring to load. |
| **RETURNS** | `Sentencizer` | The modified `Sentencizer` object. |

9
website/docs/api/textcategorizer.md
View File

@ -56,10 +56,11 @@ of problems. To handle a wider variety of problems, the `TextCategorizer` object
allows configuration of its model architecture, using the `architecture` keyword
argument.
| Name | Description |
| -------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `"ensemble"` | **Default:** Stacked ensemble of a unigram bag-of-words model and a neural network model. The neural network uses a CNN with mean pooling and attention. |
| `"simple_cnn"` | A neural network model where token vectors are calculated using a CNN. The vectors are mean pooled and used as features in a feed-forward network. |
| Name | Description |
| -------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `"ensemble"` | **Default:** Stacked ensemble of a bag-of-words model and a neural network model. The neural network uses a CNN with mean pooling and attention. The "ngram_size" and "attr" arguments can be used to configure the feature extraction for the bag-of-words model. |
| `"simple_cnn"` | A neural network model where token vectors are calculated using a CNN. The vectors are mean pooled and used as features in a feed-forward network. This architecture is usually less accurate than the ensemble, but runs faster. |
| `"bow"` | An ngram "bag-of-words" model. This architecture should run much faster than the others, but may not be as accurate, especially if texts are short. The features extracted can be controlled using the keyword arguments `ngram_size` and `attr`. For instance, `ngram_size=3` and `attr="lower"` would give lower-cased unigram, trigram and bigram features. 2, 3 or 4 are usually good choices of ngram size. |
## TextCategorizer.\_\_call\_\_ {#call tag="method"}

30
website/docs/usage/101/_architecture.md
View File

@ -25,21 +25,21 @@ an **annotated document**. It also orchestrates training and serialization.
### Processing pipeline {#architecture-pipeline}
| Name | Description |
| --------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- |
| [`Language`](/api/language) | A text-processing pipeline. Usually you'll load this once per process as `nlp` and pass the instance around your application. |
| [`Tokenizer`](/api/tokenizer) | Segment text, and create `Doc` objects with the discovered segment boundaries. |
| [`Lemmatizer`](/api/lemmatizer) | Determine the base forms of words. |
| `Morphology` | Assign linguistic features like lemmas, noun case, verb tense etc. based on the word and its part-of-speech tag. |
| [`Tagger`](/api/tagger) | Annotate part-of-speech tags on `Doc` objects. |
| [`DependencyParser`](/api/dependencyparser) | Annotate syntactic dependencies on `Doc` objects. |
| [`EntityRecognizer`](/api/entityrecognizer) | Annotate named entities, e.g. persons or products, on `Doc` objects. |
| [`TextCategorizer`](/api/textcategorizer) | Assign categories or labels to `Doc` objects. |
| [`Matcher`](/api/matcher) | Match sequences of tokens, based on pattern rules, similar to regular expressions. |
| [`PhraseMatcher`](/api/phrasematcher) | Match sequences of tokens based on phrases. |
| [`EntityRuler`](/api/entityruler) | Add entity spans to the `Doc` using token-based rules or exact phrase matches. |
| [`SentenceSegmenter`](/api/sentencesegmenter) | Implement custom sentence boundary detection logic that doesn't require the dependency parse. |
| [Other functions](/api/pipeline-functions) | Automatically apply something to the `Doc`, e.g. to merge spans of tokens. |
| Name | Description |
| ------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- |
| [`Language`](/api/language) | A text-processing pipeline. Usually you'll load this once per process as `nlp` and pass the instance around your application. |
| [`Tokenizer`](/api/tokenizer) | Segment text, and create `Doc` objects with the discovered segment boundaries. |
| [`Lemmatizer`](/api/lemmatizer) | Determine the base forms of words. |
| `Morphology` | Assign linguistic features like lemmas, noun case, verb tense etc. based on the word and its part-of-speech tag. |
| [`Tagger`](/api/tagger) | Annotate part-of-speech tags on `Doc` objects. |
| [`DependencyParser`](/api/dependencyparser) | Annotate syntactic dependencies on `Doc` objects. |
| [`EntityRecognizer`](/api/entityrecognizer) | Annotate named entities, e.g. persons or products, on `Doc` objects. |
| [`TextCategorizer`](/api/textcategorizer) | Assign categories or labels to `Doc` objects. |
| [`Matcher`](/api/matcher) | Match sequences of tokens, based on pattern rules, similar to regular expressions. |
| [`PhraseMatcher`](/api/phrasematcher) | Match sequences of tokens based on phrases. |
| [`EntityRuler`](/api/entityruler) | Add entity spans to the `Doc` using token-based rules or exact phrase matches. |
| [`Sentencizer`](/api/sentencizer) | Implement custom sentence boundary detection logic that doesn't require the dependency parse. |
| [Other functions](/api/pipeline-functions) | Automatically apply something to the `Doc`, e.g. to merge spans of tokens. |
### Other classes {#architecture-other}

109
website/docs/usage/linguistic-features.md
View File

@ -1149,9 +1149,14 @@ but it also means you'll need a **statistical model** and accurate predictions.
If your texts are closer to general-purpose news or web text, this should work
well out-of-the-box. For social media or conversational text that doesn't follow
the same rules, your application may benefit from a custom rule-based
implementation. You can either plug a rule-based component into your
[processing pipeline](/usage/processing-pipelines) or use the
`SentenceSegmenter` component with a custom strategy.
implementation. You can either use the built-in
[`Sentencizer`](/api/sentencizer) or plug an entirely custom rule-based function
into your [processing pipeline](/usage/processing-pipelines).
spaCy's dependency parser respects already set boundaries, so you can preprocess
your `Doc` using custom rules _before_ it's parsed. Depending on your text, this
may also improve accuracy, since the parser is constrained to predict parses
consistent with the sentence boundaries.
### Default: Using the dependency parse {#sbd-parser model="parser"}
@ -1168,13 +1173,35 @@ for sent in doc.sents:
print(sent.text)
```
### Setting boundaries manually {#sbd-manual}
### Rule-based pipeline component {#sbd-component}
spaCy's dependency parser respects already set boundaries, so you can preprocess
your `Doc` using custom rules _before_ it's parsed. This can be done by adding a
[custom pipeline component](/usage/processing-pipelines). Depending on your
text, this may also improve accuracy, since the parser is constrained to predict
parses consistent with the sentence boundaries.
The [`Sentencizer`](/api/sentencizer) component is a
[pipeline component](/usage/processing-pipelines) that splits sentences on
punctuation like `.`, `!` or `?`. You can plug it into your pipeline if you only
need sentence boundaries without the dependency parse.
```python
### {executable="true"}
import spacy
from spacy.lang.en import English
nlp = English() # just the language with no model
sentencizer = nlp.create_pipe("sentencizer")
nlp.add_pipe(sentencizer)
doc = nlp(u"This is a sentence. This is another sentence.")
for sent in doc.sents:
print(sent.text)
```
### Custom rule-based strategy {id="sbd-custom"}
If you want to implement your own strategy that differs from the default
rule-based approach of splitting on sentences, you can also create a
[custom pipeline component](/usage/processing-pipelines#custom-components) that
takes a `Doc` object and sets the `Token.is_sent_start` attribute on each
individual token. If set to `False`, the token is explicitly marked as _not_ the
start of a sentence. If set to `None` (default), it's treated as a missing value
and can still be overwritten by the parser.
<Infobox title="Important note" variant="warning">
@ -1187,9 +1214,11 @@ adding it to the pipeline using [`nlp.add_pipe`](/api/language#add_pipe).
Here's an example of a component that implements a pre-processing rule for
splitting on `'...'` tokens. The component is added before the parser, which is
then used to further segment the text. This approach can be useful if you want
to implement **additional** rules specific to your data, while still being able
to take advantage of dependency-based sentence segmentation.
then used to further segment the text. That's possible, because `is_sent_start`
is only set to `True` for some of the tokens – all others still specify `None`
for unset sentence boundaries. This approach can be useful if you want to
implement **additional** rules specific to your data, while still being able to
take advantage of dependency-based sentence segmentation.
```python
### {executable="true"}
@ -1212,62 +1241,6 @@ doc = nlp(text)
print("After:", [sent.text for sent in doc.sents])
```
### Rule-based pipeline component {#sbd-component}
The `sentencizer` component is a
[pipeline component](/usage/processing-pipelines) that splits sentences on
punctuation like `.`, `!` or `?`. You can plug it into your pipeline if you only
need sentence boundaries without the dependency parse. Note that `Doc.sents`
will **raise an error** if no sentence boundaries are set.
```python
### {executable="true"}
import spacy
from spacy.lang.en import English
nlp = English() # just the language with no model
sentencizer = nlp.create_pipe("sentencizer")
nlp.add_pipe(sentencizer)
doc = nlp(u"This is a sentence. This is another sentence.")
for sent in doc.sents:
print(sent.text)
```
### Custom rule-based strategy {#sbd-custom}
If you want to implement your own strategy that differs from the default
rule-based approach of splitting on sentences, you can also instantiate the
`SentenceSegmenter` directly and pass in your own strategy. The strategy should
be a function that takes a `Doc` object and yields a `Span` for each sentence.
Here's an example of a custom segmentation strategy for splitting on newlines
only:
```python
### {executable="true"}
from spacy.lang.en import English
from spacy.pipeline import SentenceSegmenter
def split_on_newlines(doc):
start = 0
seen_newline = False
for word in doc:
if seen_newline and not word.is_space:
yield doc[start:word.i]
start = word.i
seen_newline = False
elif word.text == '\\n':
seen_newline = True
if start < len(doc):
yield doc[start:len(doc)]
nlp = English() # Just the language with no model
sentencizer = SentenceSegmenter(nlp.vocab, strategy=split_on_newlines)
nlp.add_pipe(sentencizer)
doc = nlp(u"This is a sentence\\n\\nThis is another sentence\\nAnd more")
for sent in doc.sents:
print([token.text for token in sent])
```
## Rule-based matching {#rule-based-matching hidden="true"}
<div id="rule-based-matching">

2
website/docs/usage/processing-pipelines.md
View File

@ -138,7 +138,7 @@ require them in the pipeline settings in your model's `meta.json`.
| `ner` | [`EntityRecognizer`](/api/entityrecognizer) | Assign named entities. |
| `textcat` | [`TextCategorizer`](/api/textcategorizer) | Assign text categories. |
| `entity_ruler` | [`EntityRuler`](/api/entityruler) | Assign named entities based on pattern rules. |
| `sentencizer` | [`SentenceSegmenter`](/api/sentencesegmenter) | Add rule-based sentence segmentation without the dependency parse. |
| `sentencizer` | [`Sentencizer`](/api/sentencizer) | Add rule-based sentence segmentation without the dependency parse. |
| `merge_noun_chunks` | [`merge_noun_chunks`](/api/pipeline-functions#merge_noun_chunks) | Merge all noun chunks into a single token. Should be added after the tagger and parser. |
| `merge_entities` | [`merge_entities`](/api/pipeline-functions#merge_entities) | Merge all entities into a single token. Should be added after the entity recognizer. |
| `merge_subtokens` | [`merge_subtokens`](/api/pipeline-functions#merge_subtokens) | Merge subtokens predicted by the parser into single tokens. Should be added after the parser. |

2
website/docs/usage/v2-1.md
View File

@ -197,7 +197,7 @@ the existing pages and added some new content:
- **Universe:** [Videos](/universe/category/videos) and
[Podcasts](/universe/category/podcasts)
- **API:** [`EntityRuler`](/api/entityruler)
- **API:** [`SentenceSegmenter`](/api/sentencesegmenter)
- **API:** [`Sentencizer`](/api/sentencizer)
- **API:** [Pipeline functions](/api/pipeline-functions)
## Backwards incompatibilities {#incompat}

2
website/meta/sidebars.json
View File

@ -79,7 +79,7 @@
{ "text": "Matcher", "url": "/api/matcher" },
{ "text": "PhraseMatcher", "url": "/api/phrasematcher" },
{ "text": "EntityRuler", "url": "/api/entityruler" },
{ "text": "SentenceSegmenter", "url": "/api/sentencesegmenter" },
{ "text": "Sentencizer", "url": "/api/sentencizer" },
{ "text": "Other Functions", "url": "/api/pipeline-functions" }
]
},

BIN
website/src/images/icon.png
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 30 KiB

After

Width:  |  Height:  |  Size: 24 KiB