Skip to content

P
physical
Commit 5aefab77 authored by Kai's avatar Kai
Browse files
Options

update

parents 6a2b1cfe 374fb99d
Hide whitespace changes
Inline Side-by-side
Showing with 38 additions and 76 deletions
injection/data_sync/tasks.py
View file @ 5aefab77
......@@ -60,12 +60,21 @@ def sync_face_similar_data_to_redis():
item_list = list()
for item in similar_result_items:
weight_score = int(item.similarity * 100)
item_list.append(
{
"contrast_user_id": item.contrast_user_id,
"similarity": item.similarity
"filter":{
"constant_score":{
"filter":{
"term":{"user_id": item.contrast_user_id}
}
}
},
"weight": weight_score*2
}
)
if len(item_list)>=100:
break
redis_client.set(redis_key, json.dumps(item_list))
logging.info("participant_user_id:%d set data done!" % participant_user_id)
......
search/utils/topic.py
View file @ 5aefab77
......@@ -11,6 +11,8 @@ from libs.es import ESPerform
from .common import TopicDocumentField
from search.utils.common import *
from trans2es.models.pictorial import PictorialTopics
from libs.cache import redis_client
class TopicUtils(object):
......@@ -161,14 +163,6 @@ class TopicUtils(object):
q["query"] = dict()
functions_list = [
# {
# "filter": {
# "term": {
# "language_type": 1
# }
# },
# "weight": 60
# },
{
"gauss": {
"create_time": {
......@@ -177,38 +171,22 @@ class TopicUtils(object):
}
},
"weight": 60
},
# {
# "filter": {
# "constant_score":{
# "filter":{
# "term": {
# "content_level": 6
# }
# }
# }
# },
# "weight": 600
# }
}
]
# if len(user_similar_score_list) > 0:
# for item in user_similar_score_list[:100]:
# score_item = 2 + item[1]
# functions_list.append(
# {
# "filter": {"bool": {
# "should": {"term": {"user_id": item[0]}}}},
# "weight": score_item,
# }
# )
if user_id and user_id>0:
redis_key_prefix = "physical:user_similar:participant_user_id:"
similar_redis_key = redis_key_prefix + str(user_id)
redis_user_similar_data = redis_client.get(similar_redis_key)
user_similar_list = json.loads(redis_user_similar_data) if redis_user_similar_data else []
if len(user_similar_list)>0:
functions_list.extend(user_similar_list)
if len(attention_user_id_list) > 0:
functions_list.append(
{
"filter": {"bool": {
"should": {"terms": {"user_id": attention_user_id_list}}}},
"weight": 30,
"filter": {"constant_score":{"filter":{"terms": {"user_id": attention_user_id_list}}}},
"weight": 100,
}
)
if len(attention_tag_list) > 0:
......@@ -231,8 +209,6 @@ class TopicUtils(object):
"query": {
"bool": {
"filter": [
# {"term": {"content_level": 6}},
# {"term": {"has_image":True}},
{"term": {"is_online": True}},
{"term": {"is_deleted": False}}
],
......@@ -304,7 +280,7 @@ class TopicUtils(object):
]
query_function_score["query"]["bool"]["minimum_should_match"] = 1
query_function_score["query"]["bool"]["filter"].append(
{"range": {"content_level": {"gte": 4, "lte": 6}}}
{"range": {"content_level": {"gte":3,"lte":6}}}
)
else:
if "must_not" in query_function_score["query"]["bool"]:
......@@ -328,7 +304,6 @@ class TopicUtils(object):
q["collapse"] = {
"field": "user_id"
}
# "includes": ["id", "pictorial_id", "offline_score", "user_id", "edit_tag_list"]
q["_source"] = {
"includes": ["id"]
}
......
search/views/topic.py
View file @ 5aefab77
......@@ -49,13 +49,15 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list = [],disable_collpase=False):
try:
topic_star_routing = "6"
index_type = "topic-high-star"
if query is None:
if user_id>0:
redis_key = "physical:home_recommend" + ":user_id:" + str(user_id) + ":query_type:" + str(query_type)
else:
redis_key = "physical:home_recommend" + ":device_id:" + device_id + ":query_type:" + str(query_type)
else:
topic_star_routing = "4,5,6"
topic_star_routing = "3,4,5,6"
index_type = "topic"
if user_id>0:
redis_key = "physical:home_query" + ":user_id:" + str(user_id) + ":query:" + str(query) + ":query_type:" + str(query_type)
else:
......@@ -80,19 +82,10 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
# user_similar_score_redis_list = json.loads(
# redis_user_similar_score_redis_val) if redis_user_similar_score_redis_val else []
recommend_topic_user_list = list()
useful_topic_id_list = list()
attention_tag_list = list()
recommend_topic_list = list()
if query is None:
#有用标签召回
useful_topic_id_list = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=tag_id, offset=0, size=4,
single_size=size, query=query,
query_type=query_type,
filter_topic_id_list=have_read_topic_id_list,
index_type="topic-high-star",
routing=topic_star_routing,
disable_collpase=disable_collpase,usefulrecall=1)
# linucb 推荐帖子
topic_recommend_redis_key = "physical:linucb:topic_recommend:device_id:" + str(device_id)
......@@ -105,20 +98,20 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
recommend_topic_id_list.sort(key=linucb_recommend_topic_id_list.index)
# cursor = int(str(recommend_topic_dict[b"cursor"], encoding="utf-8"))
# newcursor = cursor + 6
# if len(recommend_topic_id_list) > 6:
recommend_topic_list = recommend_topic_id_list[0:8]
if len(recommend_topic_id_list) > 0:
recommend_topic_list = recommend_topic_id_list[0:size]
# redis_client.hset(topic_recommend_redis_key, "cursor", newcursor)
if b"datadict" in recommend_topic_dict:
linucb_recommend_topic_id_dict = json.loads(recommend_topic_dict[b"datadict"])
if len(recommend_topic_list) >0 and linucb_recommend_topic_id_dict is not None:
if len(recommend_topic_list) == 6 and linucb_recommend_topic_id_dict is not None:
for i in recommend_topic_list:
recommend_topic_user_list.append(linucb_recommend_topic_id_dict[str(i)])
# 用户关注标签
redis_tag_data = redis_client.hget("physical:linucb:register_user_tag_info", user_id)
attention_tag_list = json.loads(redis_tag_data) if redis_tag_data else []
if len(recommend_topic_list)>0:
size = size-len(recommend_topic_list)-len(useful_topic_id_list)
size = size-len(recommend_topic_list)
have_read_topic_id_list.extend(recommend_topic_list)
# have_read_topic_id_list_add_promote = list()
......@@ -129,11 +122,12 @@ 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)
topic_id_list = list()
rank_topic_id_list = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=tag_id, offset=0, size=size,
single_size=size,query=query, query_type=query_type,
filter_topic_id_list=have_read_topic_id_list,
index_type="topic-high-star",routing=topic_star_routing,attention_tag_list=attention_tag_list,linucb_user_id_list=recommend_topic_user_list,disable_collpase=disable_collpase)
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,
single_size=size,query=query, query_type=query_type,
filter_topic_id_list=have_read_topic_id_list,
index_type=index_type,routing=topic_star_routing,attention_tag_list=attention_tag_list,linucb_user_id_list=recommend_topic_user_list,disable_collpase=disable_collpase)
# if len(recommend_topic_list) == 6 and query is None:
# if (size < 11):
......
trans2es/models/topic.py
View file @ 5aefab77
......@@ -243,10 +243,6 @@ class Topic(models.Model):
elif user_query_results[0].is_shadow:
user_is_shadow = True
# 是否官方推荐小组
# if self.group and self.group.is_recommend:
# offline_score += 4.0
# 帖子等级
if self.content_level == '5':
offline_score += 100.0 *3
......@@ -255,20 +251,8 @@ class Topic(models.Model):
elif self.content_level == '6':
offline_score += 200.0 *3
# is_excellent = self.judge_if_excellent_topic(self.id)
# if is_excellent:
# offline_score += 200.0
if self.language_type == 1:
offline_score += 60.0
# exposure_count = ActionSumAboutTopic.objects.using(settings.SLAVE_DB_NAME).filter(topic_id=self.id, data_type=1).count()
# click_count = ActionSumAboutTopic.objects.using(settings.SLAVE_DB_NAME).filter(topic_id=self.id, data_type=2).count()
# uv_num = ActionSumAboutTopic.objects.using(settings.SLAVE_DB_NAME).filter(topic_id=self.id, data_type=3).count()
#
# if exposure_count > 0:
# offline_score += click_count / exposure_count
# if uv_num > 0:
# offline_score += (self.vote_num / uv_num + self.reply_num / uv_num)
"""
1:马甲账号是否对总分降权？
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment