load model

word_vector/word_to_vec.py

@@ -15,6 +15,10 @@ model_output_name = "w2v_model"
 model_path = os.path.join(model_dir, model_output_name)
 WORD2VEC_MODEL = word2vec.Word2Vec.load(model_path)
 
+tracate_click_ids_model_name = "tractate_click_ids_item2vec_model"
+tractate_click_ids_model_path = os.path.join(model_dir, tracate_click_ids_model_name)
+TRACTATE_CLICK_IDS_MODEL = word2vec.Word2Vec.load(tractate_click_ids_model_path)
+
 
 class W2vSentences:
     def __init__(self, f_name):
@@ -74,7 +78,7 @@ def projects_item2vec(score_limit=5):
     return model
 
 
-def clicked_tractate_ids_item2vec():
+def save_clicked_tractate_ids_item2vec():
     click_ids = []
     with open(os.path.join(data_dir, "click_tractate_ids.csv"), "r") as f:
         data = f.readlines()
@@ -86,11 +90,16 @@ def clicked_tractate_ids_item2vec():
     model = Word2Vec(click_ids, hs=0, min_count=3, workers=multiprocessing.cpu_count(), iter=10)
     print(model)
     print(len(click_ids))
-    for id in ["373744", "268517", "512"]:
-        print(model.wv.most_similar(id, topn=5))
+
+    model.save(tractate_click_ids_model_path)
     return model
 
 
+@bind("strategy_embedding/word_vector/tractate_item2vec")
+def clicked_tractate_ids_item2vec_model(id, n=5):
+    return TRACTATE_CLICK_IDS_MODEL.wv.most_similar(id, topn=n)
+
+
 if __name__ == "__main__":
     begin_time = time.time()
 
@@ -99,6 +108,9 @@ if __name__ == "__main__":
     for i in ["双眼皮", "隆鼻"]:
         print(word_similarity(i))
 
-    clicked_tractate_ids_item2vec()
+    # save_clicked_tractate_ids_item2vec()
+
+    for id in ["84375", "148764", "368399"]:
+        print(clicked_tractate_ids_item2vec_model(id, n=5))
 
     print("total cost: {:.2f}mins".format((time.time() - begin_time) / 60))