get all tags

tractate_sentence_similary.py

@@ -111,15 +111,17 @@ if __name__ == "__main__":
         words = f.readlines()
         for word in words:
             word = word.rstrip("\n")
-            all_keywords_set.add(word)
+            if word:
+                all_keywords_set.add(word)
     print("keyword: " + str(len(all_keywords_set)))
+    print(list(all_keywords_set)[:5])
 
     for word in all_keywords_set:
         jieba.add_word(word, freq=1000, tag="user_defined")
 
     # save_tractate_vector_to_redis(all_keywords_set, model)
 
-    all_tags_lst = get_all_business_tags()
+    all_tags_lst = [i.encode("utf-8") for i in get_all_business_tags()]
     print("all tags: " + str(len(all_tags_lst)))
     print(all_tags_lst[:5])
     # save_tag_vector_to_redis(all_tags_lst, model)