帖子详情页标签清洗

linucb/views/collect_data.py

@@ -173,11 +173,11 @@ class CollectData(object):
                                     click_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(
                                         topic_id=topic_id, is_online=True).values_list("tag_id", "is_collection")
                                     for tag_id, is_collection in click_results:
-                                        topic_tag_list.append(tag_id)
+                                        # topic_tag_list.append(tag_id)
                                         if is_collection:
-                                            click_topic_tag_list.append(tag_id)
+                                            topic_tag_list.append(tag_id)
                                     tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
-                                        id__in=topic_tag_list, is_online=True, is_deleted=False).values_list("id",
+                                        id__in=topic_tag_list, is_online=True, is_deleted=False, is_category=False).values_list("id",
                                                                                                              "is_ai")
                                     for id, is_ai in tag_query_results:
                                         if is_ai:
@@ -191,7 +191,7 @@ class CollectData(object):
                                     device_id = raw_val_dict["device"]["device_id"]
                                     user_id = raw_val_dict["user_id"] if "user_id" in raw_val_dict else None
                                     if query_type=="do":
-                                        tag_list = list(Tag.objects.using(settings.SLAVE_DB_NAME).filter(name=tag_name,is_online=True,is_deleted=False).values_list("id",flat=True))
+                                        tag_list = list(Tag.objects.using(settings.SLAVE_DB_NAME).filter(name=tag_name,is_online=True,is_deleted=False, is_category=False).values_list("id",flat=True))
                                         click_topic_tag_list.extend(tag_list)
                                     logging.info("query tag attention,positive tag_list,device_id:%s,query_name:%s,tag_list:%s" % (
                                     str(device_id), tag_name, str(click_topic_tag_list)))
@@ -276,15 +276,19 @@ class CollectData(object):
 
                                 # if len(exposure_sql_query_results)>0:
                                 if len(tagid_list) > 0:
+                                   tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
+                                        id__in=tagid_list, is_online=True, is_deleted=False,
+                                        is_category=False).values_list("id",flat =True)
                                    is_click = 1
                                    is_vote = 0
 
                                    reward = 1 if is_click or is_vote else 0
-                                   for tag_id in tagid_list:
+                                   for tag_id in tag_query_results:
                                        self.update_user_linucb_tag_info(reward, device_id, tag_id, user_feature)
 
                                    # 更新该用户的推荐tag数据，放在 更新完成user tag行为信息之后
-                                   self.update_recommend_tag_list(device_id, user_feature, user_id,new_user_click_tag_list=tagid_list)
+                                   self.update_recommend_tag_list(device_id, user_feature, user_id,
+                                                                  new_user_click_tag_list=tag_query_results)
                                 else:
                                     logging.warning("unknown type msg:%s" % raw_val_dict.get("type", "missing type"))
                         except:

search/views/topic.py

@@ -14,6 +14,7 @@ from search.utils.common import *
 from libs.es import ESPerform
 from django.conf import settings
 import datetime
+from trans2es.models.tag import Tag
 
 
 def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE):
@@ -390,18 +391,47 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
         have_read_topic_list.append(topic_id)
         topic_user_result = list()
         topic_tag_result = list()
+        normal_tag_result = list()
+        category_tag_reuslt = list()
         result = list()
-        if len(topic_tag_list) != 0:
-            topic_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list,
+        tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
+            id__in=topic_tag_list, is_online=True, is_deleted=False).values_list("id", "is_category")
+
+        category_tag_list = list()
+        normal_tag_list = list()
+
+        for id, is_category in tag_query_results:
+            if is_category:
+                category_tag_list.append(id)
+            else:
+                normal_tag_list.append(id)
+
+        if len(normal_tag_list) != 0:
+            normal_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list,
                                                                                size, es_cli_obj,
                                                                                index_type="topic", routing="3,4,5,6",
-                                                                              topic_tag_list=topic_tag_list)
-            topic_tag_size = len(topic_tag_result)
-            have_read_topic_list.extend(topic_tag_result)
+                                                                               topic_tag_list=normal_tag_list)
+            normal_tag_size = len(normal_tag_result)
+            have_read_topic_list.extend(normal_tag_result)
         else:
-            topic_tag_size = 0
-        if topic_tag_size < size:
-            size = size - topic_tag_size
+            normal_tag_size = 0
+
+        if normal_tag_size < size:
+            size = size - normal_tag_size
+            if len(category_tag_list) != 0:
+                category_tag_reuslt = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id,
+                                                                                     have_read_topic_list,
+                                                                                     size, es_cli_obj,
+                                                                                     index_type="topic",
+                                                                                     routing="3,4,5,6",
+                                                                                     topic_tag_list=category_tag_list)
+                category_tag_size = len(category_tag_reuslt)
+                have_read_topic_list.extend(category_tag_reuslt)
+            else:
+                category_tag_size = 0
+
+            if category_tag_size < size:
+                size = size - category_tag_size
                 if topic_user_id != -1:
                     topic_user_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id,
                                                                                    have_read_topic_list,
@@ -454,7 +484,8 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
         logging.warning("topic_user_result:%s" % str(topic_user_result))
         logging.warning("result:%s" % str(result))
         recommend_topic_ids_list = list()
-        recommend_topic_ids_list.extend(topic_tag_result)
+        recommend_topic_ids_list.extend(normal_tag_result)
+        recommend_topic_ids_list.extend(category_tag_reuslt)
         recommend_topic_ids_list.extend(topic_user_result)
         recommend_topic_ids_list.extend(result)
         return {"recommend_topic_ids": recommend_topic_ids_list}