调试

linucb/core/Linucb.py

@@ -21,8 +21,8 @@ import time
 
 
 class LinUCB:
-    d = 3
-    alpha = 10
+    d = 2
+    alpha = 0.1
     r1 = 10
     r0 = -0.1
     default_tag_list = list()
@@ -170,7 +170,7 @@ class LinUCB:
             return False
 
     @classmethod
-    def update_linucb_info(cls, user_features,reward, content_id, redis_name_linucb_matrix,redis_cli):
+    def update_linucb_info(cls,reward, content_id, redis_name_linucb_matrix,redis_cli,user_features=None,user_features_array=None):
         try:
             if reward == -1:
                 logging.warning("reward val error!")
@@ -180,7 +180,10 @@ class LinUCB:
                 else:
                     r = cls.r0
 
-                xaT = np.array([user_features])
+                if user_features_array:
+                    xaT = np.array(user_features_array)
+                else:
+                    xaT = np.array([user_features])
                 xa = np.transpose(xaT)
 
                 ori_redis_tag_data = redis_cli.hget(redis_name_linucb_matrix, content_id)

tools/generate_content_user_feature.py

@@ -85,7 +85,7 @@ class Generate_Feature_Info(object):
 
 
     @classmethod
-    def generate_content_feature_to_redis(cls,label_encoder,content_type="diary"):
+    def generate_content_feature_to_redis(cls,onehot_encoder,content_type="diary"):
         try:
             redis_name_content_linucb_feature = cls.redis_name_content_linucb_feature_prefix + content_type
 
@@ -138,12 +138,11 @@ class Generate_Feature_Info(object):
                         tag_label = -1
                         if len(offi_tags_v3)>0:
                             try:
-                                tag_label = label_encoder.transform(offi_tags_v3)[0]
+                                diary_feature_list = onehot_encoder.transform([offi_tags_v3]).toarray().tolist()
+                                redis_client.hset(redis_name_content_linucb_feature, diary_id,
+                                              json.dumps(diary_feature_list))
                             except:
-                                pass
-
-                        diary_feature_list = [tag_label,content_level]
-                        redis_client.hset(redis_name_content_linucb_feature,diary_id,json.dumps(diary_feature_list))
+                                logging.error("catch exception,err_msg:%s" % traceback.format_exc())
 
                 except:
                     logging.error("catch exception,err_msg:%s" % traceback.format_exc())
@@ -157,7 +156,7 @@ 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,label_encoder):
+    def generate_user_feature_to_redis(cls,device_id,onehot_encoder):
         try:
             redis_name_user_linucb_feature = cls.redis_name_user_linucb_feature_prefix
 
@@ -170,15 +169,10 @@ 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_tag_label = label_encoder.transform([user_max_score_tag])[0]
-                    print(user_max_score_tag)
-                    print(user_tag_label)
+                    user_feature_list = onehot_encoder.transform([[user_max_score_tag]]).toarray().tolist()
+                    redis_client.hset(redis_name_user_linucb_feature, device_id, json.dumps(user_feature_list))
                 except:
-                    pass
-
-            user_feature_list = [user_tag_label]
-            redis_client.hset(redis_name_user_linucb_feature,device_id,json.dumps(user_feature_list))
-
+                    logging.error("catch exception,err_msg:%s" % traceback.format_exc())
         except:
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
 
@@ -191,7 +185,7 @@ class Generate_Feature_Info(object):
 
             user_feature_list = list()
             if user_feature_redis_data:
-                user_feature_list = [float(json.loads(user_feature_redis_data)[0])]
+                user_feature_list = json.loads(user_feature_redis_data)
 
             return user_feature_list
         except:
@@ -206,8 +200,7 @@ class Generate_Feature_Info(object):
 
             content_feature_redis_data = redis_client.hget(redis_name_content_linucb_feature, card_id)
             if content_feature_redis_data:
-                content_feature_redis_list = json.loads(content_feature_redis_data)
-                user_feature_list = [float(item) for item in content_feature_redis_list]
+                user_feature_list = json.loads(content_feature_redis_data)
 
             return user_feature_list
         except:
@@ -225,7 +218,7 @@ 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(label_encoder)
+    Generate_Feature_Info.generate_content_feature_to_redis(onehot_encoder)
 
 
     linucb_matrix_redis_name = "strategy:linucb:matrix:content_type:diary"
@@ -242,15 +235,18 @@ 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, label_encoder)
+        Generate_Feature_Info.generate_user_feature_to_redis(device_id, 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)
 
-        user_feature_list = user_feature_list+content_feature_list
-        print(user_feature_list)
+        offi_context_feature_list = list()
+        if len(user_feature_list)>0 and len(content_feature_list)>0:
+            offi_context_feature_list.append(user_feature_list[0])
+            offi_context_feature_list.append(content_feature_list[0])
 
-        LinUCB.update_linucb_info(user_features=user_feature_list,reward=1,content_id=diary_id,
+        print(offi_context_feature_list)
+        LinUCB.update_linucb_info(user_features_array=offi_context_feature_list,reward=1,content_id=diary_id,
                                   redis_name_linucb_matrix=linucb_matrix_redis_name,redis_cli=redis_client)
     diary_fd.close()
     # test_val_list = ["切开双眼皮"]