update

eda/smart_rank/dist_update_user_portrait.py

@@ -41,7 +41,7 @@ def compute_validate(x):
         return score
     else:
         return 0.5
-def tag_list2dict(lst):
+def tag_list2dict(lst, size):
     result = []
     if lst:
         for i in lst:
@@ -53,10 +53,10 @@ def tag_list2dict(lst):
                 tmp["type"] = "search_word"
             tmp["score"] = i["tag_score"]
             result.append(tmp)
-    return result
+    return result[:size]
 
 
-def get_user_tag_score(cl_id, all_log_df):
+def get_user_tag_score(cl_id, all_log_df, size=10):
     user_log_df = all_log_df.loc[all_log_df['cl_id'] == cl_id]
     if not user_log_df.empty:
         user_log_df["tag_id"] = np.where(user_log_df["action"] == "do_search",user_log_df["tag_referrer"],user_log_df["tag_id"])
@@ -68,7 +68,7 @@ def get_user_tag_score(cl_id, all_log_df):
         finally_score = user_log_df.sort_values(by=["tag_score","time"],ascending=False)
         finally_score.drop_duplicates(subset="tag_id", inplace=True)
         finally_score_lst = finally_score[["tag_id","tag_score"]].to_dict('record')
-        tag_id_list = tag_list2dict(finally_score_lst)
+        tag_id_list = tag_list2dict(finally_score_lst, size)
         return cl_id, tag_id_list
     else:
         return ()