most_similar_cosmul bug fix

gensim/models/keyedvectors.py

@@ -455,14 +455,21 @@ def most_similar_cosmul(self, positive=[], negative=[], topn=10):
             # allow calls like most_similar_cosmul('dog'), as a shorthand for most_similar_cosmul(['dog'])
             positive = [positive]
 
+        all_words = set([self.vocab[word].index for word in positive+negative
+            if not isinstance(word, ndarray) and word in self.vocab])
+
+        positive = [
+            self.word_vec(word, use_norm=True) if isinstance(word, string_types) else word
+            for word in positive
+        ]
+        negative = [
+            self.word_vec(word, use_norm=True) if isinstance(word, string_types) else word
+            for word in negative
+        ]
 
-        positive = [self.word_vec(word, use_norm=True) for word in positive]
-        negative = [self.word_vec(word, use_norm=True) for word in negative]
         if not positive:
             raise ValueError("cannot compute similarity with no input")
 
-        all_words = set([self.vocab[word].index for word in positive+negative if word in self.vocab])
-
         # equation (4) of Levy & Goldberg "Linguistic Regularities...",
         # with distances shifted to [0,1] per footnote (7)
         pos_dists = [((1 + dot(self.syn0norm, term)) / 2) for term in positive]

gensim/test/test_fasttext_wrapper.py

@@ -158,11 +158,11 @@ def testMostSimilar(self):
     def testMostSimilarCosmul(self):
         """Test most_similar_cosmul for in-vocab and out-of-vocab words"""
         # In vocab, sanity check
-        self.assertEqual(len(self.test_model.most_similar(positive=['the', 'and'], topn=5)), 5)
-        self.assertEqual(self.test_model.most_similar('the'), self.test_model.most_similar(positive=['the']))
+        self.assertEqual(len(self.test_model.most_similar_cosmul(positive=['the', 'and'], topn=5)), 5)
+        self.assertEqual(self.test_model.most_similar_cosmul('the'), self.test_model.most_similar_cosmul(positive=['the']))
         # Out of vocab check
-        self.assertEqual(len(self.test_model.most_similar(['night', 'nights'], topn=5)), 5)
-        self.assertEqual(self.test_model.most_similar('nights'), self.test_model.most_similar(positive=['nights']))
+        self.assertEqual(len(self.test_model.most_similar_cosmul(['night', 'nights'], topn=5)), 5)
+        self.assertEqual(self.test_model.most_similar_cosmul('nights'), self.test_model.most_similar_cosmul(positive=['nights']))
 
     def testLookup(self):
         """Tests word vector lookup for in-vocab and out-of-vocab words"""
@@ -218,4 +218,4 @@ def testHash(self):
 
 if __name__ == '__main__':
     logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.DEBUG)
-    unittest.main()
+    unittest.main()

gensim/test/test_word2vec.py

@@ -449,6 +449,17 @@ def test_cbow_neg(self):
                                   min_count=5, iter=10, workers=2, sample=0)
         self.model_sanity(model)
 
+    def test_cosmul(self):
+        model = word2vec.Word2Vec(sentences, size=2, min_count=1, hs=1, negative=0)
+        sims = model.most_similar_cosmul('graph', topn=10)
+        # self.assertTrue(sims[0][0] == 'trees', sims)  # most similar
+
+        # test querying for "most similar" by vector
+        graph_vector = model.wv.syn0norm[model.wv.vocab['graph'].index]
+        sims2 = model.most_similar_cosmul(positive=[graph_vector], topn=11)
+        sims2 = [(w, sim) for w, sim in sims2 if w != 'graph']  # ignore 'graph' itself
+        self.assertEqual(sims, sims2)
+
     def testTrainingCbow(self):
         """Test CBOW word2vec training."""
         # to test training, make the corpus larger by repeating its sentences over and over