Merge branch 'hk_now' into 'test'

add useful tag recall

See merge request alpha/physical!339

search/utils/topic.py

@@ -333,6 +333,74 @@ class TopicUtils(object):
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
             return list()
 
+    @classmethod
+    def userful_tag_topic_list(cls, user_id, have_read_topic_list, size,
+                                            index_type="topic-high-star",routing=None):
+        """
+        :remark 帖子详情页推荐列表,缺少按时间衰减
+        :param user_id:
+        :param topic_tag_list:
+        :param topic_group_id:
+        :param topic_user_id:
+        :param offset:
+        :param size:
+        :return:
+        """
+        try:
+            es_cli_obj = ESPerform.get_cli()
+            useful_tag_list = list()
+            q = dict()
+            q["query"] = {
+                "term": {
+                    "user_id": user_id
+                }
+            }
+
+            q["_source"] = {
+                "include": ["useful_tag_list"]
+            }
+            result_dict = ESPerform.get_search_results(es_cli_obj, "user", q, 0, 1)
+            if len(result_dict["hits"]) == 0:
+                logging.warning("not find user_id:%d in es!" % int(user_id))
+            else:
+                useful_tag_list = result_dict["hits"][0]["_source"]["useful_tag_list"]
+            if len(useful_tag_list) == 0:
+                return []
+            else:
+                q = dict()
+                q["query"] = dict()
+                # logging.warning("topic_tag_list:%s"%str(topic_tag_list))
+                query_function_score = {
+                        "query": {
+                            "bool": {
+                                "must": [
+                                    {"term": {"is_online": True}},
+                                    {"term": {"is_deleted": False}},
+                                    {"terms": {"tag_list": useful_tag_list}}
+                                ],
+                                "must_not": {
+                                    "terms": {
+                                        "id": have_read_topic_list
+                                    }
+                                }
+                            }
+                        }
+                   }
+                q["_source"] = {
+                    "includes": ["id"]
+                }
+
+                result_dict = ESPerform.get_search_results(es_cli_obj, sub_index_name=index_type, query_body=q, size=size,
+                                                           routing=routing)
+                topic_id_list = list()
+
+                for item in result_dict["hits"]:
+                    topic_id_list.append(item["_source"]["id"])
+                return topic_id_list
+        except:
+            logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+            return []
+
     @classmethod
     def get_topic_detail_recommend_list(cls, user_id, topic_id, topic_tag_list, topic_pictorial_id, topic_user_id,
                                         filter_topic_user_id, have_read_topic_list, offset, size, es_cli_obj=None,

search/views/topic.py

@@ -46,7 +46,7 @@ def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageTy
 
 
 def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query=None,
-                                 query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list = [],disable_collpase=False):
+                                 query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list = [],disable_collpase=False,usefulrecall = -1):
     try:
         topic_star_routing = "6"
         index_type = "topic-high-star"
@@ -76,13 +76,18 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
             else:
                 if offset>0: # 首次搜索时不需要过滤已读
                     have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
-
-        # user_similar_score_redis_key = "physical:user_similar_score:user_id:" + str(user_id)
-        # redis_user_similar_score_redis_val = redis_client.get(user_similar_score_redis_key)
-        # user_similar_score_redis_list = json.loads(
-        #     redis_user_similar_score_redis_val) if redis_user_similar_score_redis_val else []
+        have_read_topic_id_list.extend(promote_topic_list)
+        useful_topic_id_list = list()
+        #有用标签召回
+        if usefulrecall != -1:
+            useful_topic_id_list = TopicUtils.userful_tag_topic_list(user_id,have_read_topic_id_list,4,"topic-high-star","6")
+            # user_similar_score_redis_key = "physical:user_similar_score:user_id:" + str(user_id)
+            # redis_user_similar_score_redis_val = redis_client.get(user_similar_score_redis_key)
+            # user_similar_score_redis_list = json.loads(
+            #     redis_user_similar_score_redis_val) if redis_user_similar_score_redis_val else []
+            size = size - len(useful_topic_id_list)
+            have_read_topic_id_list = have_read_topic_id_list.extend(useful_topic_id_list)
         recommend_topic_user_list = list()
-
         attention_tag_list = list()
         recommend_topic_list = list()
         if query is None:
@@ -121,7 +126,6 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
             #
             # for topic_id in promote_recommend_topic_id_list:
             #     have_read_topic_id_list_add_promote.append(topic_id)
-        have_read_topic_id_list.extend(promote_topic_list)
         rank_topic_id_list = list()
         if size>0:
             rank_topic_id_list = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=tag_id, offset=0, size=size,
@@ -154,7 +158,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
         # 每个session key保存60天
         redis_client.expire(redis_key, 60 * 60 * 24 * 60)
 
-        return recommend_topic_list,rank_topic_id_list
+        return recommend_topic_list,rank_topic_id_list,useful_topic_id_list
     except:
         logging.error("catch exception,err_msg:%s" % traceback.format_exc())
         return [],[]
@@ -170,7 +174,7 @@ def query_tag_id_by_topic(offset=0, size=10, topic_id_list=[], user_id=-1):
 
 
 @bind("physical/search/home_recommend")
-def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list=[]):
+def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list=[],usefulrecall = -1):
     """
     :remark:首页推荐，目前只推荐日记
     :param session_id:
@@ -187,6 +191,7 @@ def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=Topic
 
         recommend_topic_ids = list()
         rank_topic_ids = list()
+        useful_topic_ids = list()
         es_node_load_high_flag = False
         # try:
         #     es_node_load_high_flag = ESPerform.if_es_node_load_high(ESPerform.get_cli())
@@ -207,13 +212,18 @@ def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=Topic
                     recommend_topic_ids = have_read_topic_id_list[0:size]
 
         else:
-            recommend_topic_ids,rank_topic_ids = get_home_recommend_topic_ids(user_id, device_id, tag_id=0, offset=0, size=size,
-                                                           query_type=query_type,promote_topic_list=promote_topic_list)
-
-        return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids}
+            recommend_topic_ids,rank_topic_ids,useful_topic_ids = get_home_recommend_topic_ids(user_id, device_id, tag_id=0, offset=0, size=size,
+                                                           query_type=query_type,promote_topic_list=promote_topic_list,usefulrecall=usefulrecall)
+        if usefulrecall != -1:
+            return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids,"useful_topic_ids":useful_topic_ids}
+        else:
+            return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids}
     except:
         logging.error("catch exception,err_msg:%s" % traceback.format_exc())
-        return {"linucb_topic_ids": [],"rank_topic_ids":[]}
+        if usefulrecall != -1:
+          return {"linucb_topic_ids": [],"rank_topic_ids":[],"useful_topic_ids":[]}
+        else:
+          return {"linucb_topic_ids": [], "rank_topic_ids": []}
 
 
 @bind("physical/search/discover_page")