首页feed增加低质量日记排序，增加offline_score_v1灰度实验

eda/smart_rank/gm_feed_cold_start.py

@@ -12,7 +12,7 @@ def get_all_city_id():
     return city_tag_ids
 
 
-def search_diary_by_match_phrase(tag_names, city_tag_id):
+def search_diary_by_match_phrase(tag_names, city_tag_id, version):
     q = dict()
     if city_tag_id == -1:
         sort_list = []
@@ -29,12 +29,22 @@ def search_diary_by_match_phrase(tag_names, city_tag_id):
                 '_cache': True,
             }}
         ]
-    sort_list += [
-        {'has_video_cover': {'order': 'asc'}},
-        {"good_click": {"order": "desc"}},
-        {'offline_score': {'order': 'desc'}},
-        {'last_update_time': {'order': 'desc'}}
-    ]
+    if version:
+        sort_list += [
+            {'is_low_quality': {'order': 'asc'}},
+            {'has_video_cover': {'order': 'asc'}},
+            {"good_click": {"order": "desc"}},
+            {'offline_score_v1': {'order': 'desc'}},
+            {'last_update_time': {'order': 'desc'}}
+        ]
+    else:
+        sort_list += [
+            {'is_low_quality': {'order': 'asc'}},
+            {'has_video_cover': {'order': 'asc'}},
+            {"good_click": {"order": "desc"}},
+            {'offline_score': {'order': 'desc'}},
+            {'last_update_time': {'order': 'desc'}}
+        ]
     total_query_should_list = []
     for tag_name in tag_names:
         term_dict = {
@@ -148,6 +158,7 @@ if __name__ == "__main__":
     light_clinic_beauty_topic_queue_key = "coldstart:light:clinic:beauty:topic:queue"
     light_clinic_beauty_qa_queue_key = "coldstart:light:clinic:beauty:qa:queue"
     light_clinic_beauty_diary_queue_key = "coldstart:light:clinic:beauty:diary:queue"
+    light_clinic_beauty_diary_queue_key2 = "coldstart:light:clinic:beauty:diary:queue:offline:score:v1"
     card_types = ['topic', 'qa']  # 日记有400多个城市，且是hmset，因此不用重置
     # word_refers = ['coldstart:hot:search:word', 'coldstart:light:clinic:beauty']
     word_refers = ['coldstart:light:clinic:beauty']
@@ -195,7 +206,14 @@ if __name__ == "__main__":
     print("轻医美词更新的问答队列长度：%s" % str(len(light_clinic_beauty_qa_queue)))
 
     for city_tag_id in all_city_tag_id:
-        diary_queue = search_diary_by_match_phrase(light_clinic_beauty, city_tag_id)
+        diary_queue = search_diary_by_match_phrase(light_clinic_beauty, city_tag_id, version=0)
         light_clinic_beauty_diary_queue.update({city_tag_id: json.dumps(diary_queue)})
     redis_client.hmset(light_clinic_beauty_diary_queue_key, light_clinic_beauty_diary_queue)
     print("轻医美词更新的日记队列长度：%s" % str(len(diary_queue)))
+
+    light_clinic_beauty_diary_queue = dict()
+    for city_tag_id in all_city_tag_id:
+        diary_queue = search_diary_by_match_phrase(light_clinic_beauty, city_tag_id, version=1)
+        light_clinic_beauty_diary_queue.update({city_tag_id: json.dumps(diary_queue)})
+    redis_client.hmset(light_clinic_beauty_diary_queue_key2, light_clinic_beauty_diary_queue)
+    print("轻医美词更新的日记队列长度：%s" % str(len(diary_queue)))