调试

tools/generate_content_user_feature.py

@@ -85,7 +85,7 @@ class Generate_Feature_Info(object):
 
 
     @classmethod
-    def generate_content_feature_to_redis(cls,onehot_encoder,content_type="diary"):
+    def generate_content_feature_to_redis(cls,label_encoder,onehot_encoder,content_type="diary"):
         try:
             redis_name_content_linucb_feature = cls.redis_name_content_linucb_feature_prefix + content_type
 
@@ -138,7 +138,8 @@ class Generate_Feature_Info(object):
                         tag_label = -1
                         if len(offi_tags_v3)>0:
                             try:
-                                diary_feature_list = onehot_encoder.transform([offi_tags_v3]).toarray().tolist()
+                                tag_label = label_encoder.transform(offi_tags_v3)[0]
+                                diary_feature_list = onehot_encoder.transform([tag_label]).toarray().tolist()
                                 redis_client.hset(redis_name_content_linucb_feature, diary_id,
                                               json.dumps(diary_feature_list))
                             except:
@@ -156,7 +157,7 @@ class Generate_Feature_Info(object):
         return "doris:user_portrait:tag3:device_id:" + str(device_id)
 
     @classmethod
-    def generate_user_feature_to_redis(cls,device_id,onehot_encoder):
+    def generate_user_feature_to_redis(cls,device_id,label_encoder,onehot_encoder):
         try:
             redis_name_user_linucb_feature = cls.redis_name_user_linucb_feature_prefix
 
@@ -169,7 +170,8 @@ class Generate_Feature_Info(object):
                 sorted_user_portrait_list = sorted(user_portrait_dict["projects"].items(), key=lambda x: x[1], reverse=True)
                 user_max_score_tag = sorted_user_portrait_list[0][0].encode("utf-8")
                 try:
-                    user_feature_list = onehot_encoder.transform([[user_max_score_tag]]).toarray().tolist()
+                    user_tag_label = label_encoder.transform([user_max_score_tag])[0]
+                    user_feature_list = onehot_encoder.transform([[user_tag_label]]).toarray().tolist()
                     redis_client.hset(redis_name_user_linucb_feature, device_id, json.dumps(user_feature_list))
                 except:
                     logging.error("catch exception,err_msg:%s" % traceback.format_exc())
@@ -218,7 +220,7 @@ if __name__ == "__main__":
 
     # device_id="868771031984211"
     # Generate_Feature_Info.generate_user_feature_to_redis(device_id,label_encoder)
-    Generate_Feature_Info.generate_content_feature_to_redis(onehot_encoder)
+    Generate_Feature_Info.generate_content_feature_to_redis(label_encoder,onehot_encoder)
 
 
     linucb_matrix_redis_name = "strategy:linucb:matrix:content_type:diary"
@@ -235,7 +237,7 @@ if __name__ == "__main__":
         item_list = line.split(",")
         device_id = item_list[3]
         diary_id = item_list[4]
-        Generate_Feature_Info.generate_user_feature_to_redis(device_id, onehot_encoder)
+        Generate_Feature_Info.generate_user_feature_to_redis(device_id, label_encoder,onehot_encoder)
 
         user_feature_list = Generate_Feature_Info.get_user_feature_by_device_id(device_id)
         content_feature_list = Generate_Feature_Info.get_content_feature(diary_id)