Merge branch 'master' of https://github.com/honnibal/spaCy

spacy/en/__init__.py

@@ -6,7 +6,39 @@ from ..language import Language
 
 LOCAL_DATA_DIR = path.join(path.dirname(__file__), 'data')
 
+
+# improved list from Stone, Denis, Kwantes (2010)
+STOPWORDS = """
+a about above across after afterwards again against all almost alone along already also although always am among amongst amoungst amount an and another any anyhow anyone anything anyway anywhere are around as at back be
+became because become becomes becoming been before beforehand behind being below beside besides between beyond bill both bottom but by call can
+cannot cant co computer con could couldnt cry de describe
+detail did didn do does doesn doing don done down due during
+each eg eight either eleven else elsewhere empty enough etc even ever every everyone everything everywhere except few fifteen
+fify fill find fire first five for former formerly forty found four from front full further get give go
+had has hasnt have he hence her here hereafter hereby herein hereupon hers herself him himself his how however hundred i ie
+if in inc indeed interest into is it its itself keep last latter latterly least less ltd
+just
+kg km
+made make many may me meanwhile might mill mine more moreover most mostly move much must my myself name namely
+neither never nevertheless next nine no nobody none noone nor not nothing now nowhere of off
+often on once one only onto or other others otherwise our ours ourselves out over own part per
+perhaps please put rather re
+quite
+rather really regarding
+same say see seem seemed seeming seems serious several she should show side since sincere six sixty so some somehow someone something sometime sometimes somewhere still such system take ten
+than that the their them themselves then thence there thereafter thereby therefore therein thereupon these they thick thin third this those though three through throughout thru thus to together too top toward towards twelve twenty two un under
+until up unless upon us used using
+various very very via
+was we well were what whatever when whence whenever where whereafter whereas whereby wherein whereupon wherever whether which while whither who whoever whole whom whose why will with within without would yet you
+your yours yourself yourselves
+"""
+STOPWORDS = set(w for w in STOPWORDS.split() if w)
+
 class English(Language):
     @classmethod
     def default_data_dir(cls):
         return LOCAL_DATA_DIR
+
+    @staticmethod
+    def is_stop(string):
+        return 1 if string.lower() in STOPWORDS else 0

spacy/language.py

@@ -94,6 +94,10 @@ class Language(object):
     def like_email(string):
         return orth.like_email(string)
 
+    @staticmethod
+    def is_stop(string):
+        return 0
+
     @classmethod
     def default_lex_attrs(cls, data_dir=None):
         return {
@@ -116,7 +120,7 @@ class Language(object):
             attrs.LIKE_URL: cls.like_url,
             attrs.LIKE_NUM: cls.like_number,
             attrs.LIKE_EMAIL: cls.like_email,
-            attrs.IS_STOP: lambda string: False,
+            attrs.IS_STOP: cls.is_stop,
             attrs.IS_OOV: lambda string: True
         }
 

spacy/lexeme.pyx

@@ -3,6 +3,13 @@ from cpython.ref cimport Py_INCREF
 from cymem.cymem cimport Pool
 from murmurhash.mrmr cimport hash64
 
+# Compiler crashes on memory view coercion without this. Should report bug.
+from cython.view cimport array as cvarray
+cimport numpy as np
+np.import_array()
+
+
+
 from libc.string cimport memset
 
 from .orth cimport word_shape
@@ -35,6 +42,26 @@ cdef class Lexeme:
     def py_check_flag(self, attr_id_t flag_id):
         return True if Lexeme.check_flag(self.c, flag_id) else False
 
+    def similarity(self, other):
+        return numpy.dot(self.vector, other.vector) / (self.vector_norm * other.vector_norm)
+
+    property vector_norm:
+        def __get__(self):
+            return self.c.l2_norm
+
+        def __set__(self, float value):
+            self.c.l2_norm = value
+
+    property vector:
+        def __get__(self):
+            cdef int length = self.vocab.repvec_length
+            repvec_view = <float[:length,]>self.c.repvec
+            return numpy.asarray(repvec_view)
+
+    property repvec:
+        def __get__(self):
+            return self.vector
+        
     property orth_:
         def __get__(self):
             return self.vocab.strings[self.c.orth]

spacy/tokens/doc.pyx

@@ -3,7 +3,9 @@ from libc.string cimport memcpy, memset
 from libc.stdint cimport uint32_t
 
 import numpy
+import numpy.linalg
 import struct
+cimport numpy as np
 
 from ..lexeme cimport Lexeme
 from ..lexeme cimport EMPTY_LEXEME
@@ -118,6 +120,22 @@ cdef class Doc:
     def __str__(self):
         return u''.join([t.string for t in self])
 
+    def similarity(self, other):
+        return numpy.dot(self.vector, other.vector) / (self.vector_norm * other.vector_norm)
+
+    property repvec:
+        def __get__(self):
+            return self.vector
+
+    property vector:
+        def __get__(self):
+            return sum(t.vector for t in self if not t.is_stop) / len(self)
+
+
+    property vector_norm:
+        def __get__(self):
+            return numpy.linalg.norm(self.vector)
+
     @property
     def string(self):
         return u''.join([t.string for t in self])

spacy/tokens/spans.pyx

@@ -1,5 +1,8 @@
 from __future__ import unicode_literals
 from collections import defaultdict
+import numpy
+import numpy.linalg
+cimport numpy as np
 
 from ..structs cimport TokenC, LexemeC
 from ..typedefs cimport flags_t, attr_t
@@ -52,6 +55,17 @@ cdef class Span:
     def merge(self, unicode tag, unicode lemma, unicode ent_type):
         self._seq.merge(self[0].idx, self[-1].idx + len(self[-1]), tag, lemma, ent_type)
 
+    def similarity(self, other):
+        return numpy.dot(self.vector, other.vector) / (self.vector_norm * other.vector_norm)
+
+    property vector:
+        def __get__(self):
+            return sum(t.vector for t in self if not t.is_stop) / len(self)
+
+    property vector_norm:
+        def __get__(self):
+            return numpy.linalg.norm(self.vector)
+
     property text:
         def __get__(self):
             return u' '.join([t.text for t in self])

spacy/tokens/token.pyx

@@ -49,6 +49,9 @@ cdef class Token:
     def nbor(self, int i=1):
         return self.doc[self.i+i]
 
+    def similarity(self, other):
+        return numpy.dot(self.vector, other.vector) / (self.vector_norm * other.vector_norm)
+
     property lex_id:
         def __get__(self):
             return self.c.lex.id
@@ -125,12 +128,20 @@ cdef class Token:
         def __get__(self):
             return self.c.dep
 
-    property repvec:
+    property vector:
         def __get__(self):
             cdef int length = self.vocab.repvec_length
             repvec_view = <float[:length,]>self.c.lex.repvec
             return numpy.asarray(repvec_view)
 
+    property repvec:
+        def __get__(self):
+            return self.vector
+
+    property vector_norm:
+        def __get__(self):
+            return self.c.lex.l2_norm
+
     property n_lefts:
         def __get__(self):
             cdef int n = 0
@@ -302,6 +313,9 @@ cdef class Token:
     property is_oov:
         def __get__(self): return Lexeme.check_flag(self.c.lex, IS_OOV)
 
+    property is_stop:
+        def __get__(self): return Lexeme.check_flag(self.c.lex, IS_STOP)
+
     property is_alpha:
         def __get__(self): return Lexeme.check_flag(self.c.lex, IS_ALPHA)
 

tests/vectors/test_vectors.py

@@ -0,0 +1,109 @@
+import pytest
+
+@pytest.mark.models
+def test_token_vector(EN):
+    token = EN(u'Apples and oranges')[0]
+    token.vector
+    token.vector_norm
+
+@pytest.mark.models
+def test_lexeme_vector(EN):
+    lexeme = EN.vocab[u'apples']
+    lexeme.vector
+    lexeme.vector_norm
+
+
+@pytest.mark.models
+def test_doc_vector(EN):
+    doc = EN(u'Apples and oranges')
+    doc.vector
+    doc.vector_norm
+
+@pytest.mark.models
+def test_span_vector(EN):
+    span = EN(u'Apples and oranges')[0:2]
+    span.vector
+    span.vector_norm
+
+@pytest.mark.models
+def test_token_token_similarity(EN):
+    apples, oranges = EN(u'apples oranges')
+    assert apples.similarity(oranges) == oranges.similarity(apples)
+    assert 0.0 < apples.similarity(oranges) < 1.0
+    
+
+@pytest.mark.models
+def test_token_lexeme_similarity(EN):
+    apples = EN(u'apples')
+    oranges = EN.vocab[u'oranges']
+    assert apples.similarity(oranges) == oranges.similarity(apples)
+    assert 0.0 < apples.similarity(oranges) < 1.0
+ 
+
+@pytest.mark.models
+def test_token_span_similarity(EN):
+    doc = EN(u'apples orange juice')
+    apples = doc[0]
+    oranges = doc[1:3]
+    assert apples.similarity(oranges) == oranges.similarity(apples)
+    assert 0.0 < apples.similarity(oranges) < 1.0
+ 
+
+@pytest.mark.models
+def test_token_doc_similarity(EN):
+    doc = EN(u'apples orange juice')
+    apples = doc[0]
+    assert apples.similarity(doc) == doc.similarity(apples)
+    assert 0.0 < apples.similarity(doc) < 1.0
+ 
+
+@pytest.mark.models
+def test_lexeme_span_similarity(EN):
+    doc = EN(u'apples orange juice')
+    apples = EN.vocab[u'apples']
+    span = doc[1:3]
+    assert apples.similarity(span) == span.similarity(apples)
+    assert 0.0 < apples.similarity(span) < 1.0
+
+
+@pytest.mark.models
+def test_lexeme_lexeme_similarity(EN):
+    apples = EN.vocab[u'apples']
+    oranges = EN.vocab[u'oranges']
+    assert apples.similarity(oranges) == oranges.similarity(apples)
+    assert 0.0 < apples.similarity(oranges) < 1.0
+ 
+
+@pytest.mark.models
+def test_lexeme_doc_similarity(EN):
+    doc = EN(u'apples orange juice')
+    apples = EN.vocab[u'apples']
+    assert apples.similarity(doc) == doc.similarity(apples)
+    assert 0.0 < apples.similarity(doc) < 1.0
+ 
+
+@pytest.mark.models
+def test_span_span_similarity(EN):
+    doc = EN(u'apples orange juice')
+    apples = doc[0:2]
+    oj = doc[1:3]
+    assert apples.similarity(oj) == oj.similarity(apples)
+    assert 0.0 < apples.similarity(oj) < 1.0
+ 
+
+@pytest.mark.models
+def test_span_doc_similarity(EN):
+    doc = EN(u'apples orange juice')
+    apples = doc[0:2]
+    oj = doc[1:3]
+    assert apples.similarity(doc) == doc.similarity(apples)
+    assert 0.0 < apples.similarity(doc) < 1.0
+ 
+
+@pytest.mark.models
+def test_doc_doc_similarity(EN):
+    apples = EN(u'apples and apple pie')
+    oranges = EN(u'orange juice')
+    assert apples.similarity(oranges) == apples.similarity(oranges)
+    assert 0.0 < apples.similarity(oranges) < 1.0
+