|
@@ -9,17 +9,17 @@ with open("article_topics") as f:
|
9
|
9
|
lines=f.readlines()
|
10
|
10
|
lines=[line.strip() for line in lines]
|
11
|
11
|
for line in lines:
|
12
|
|
- data=json.loads(line)
|
13
|
|
- url=data["__url"]
|
14
|
|
- aid=data["article_id"]
|
15
|
|
- feature=[0 for _ in range(0,length)]
|
16
|
|
- index_url[aid]=url
|
17
|
|
- for index,score in data["topic_lda"]:
|
18
|
|
- feature[index]=score
|
19
|
|
- t.add_item(aid, feature)
|
|
12
|
+ data=json.loads(line)
|
|
13
|
+ url=data["__url"]
|
|
14
|
+ aid=data["article_id"]
|
|
15
|
+ feature=[0 for _ in range(0,length)]
|
|
16
|
+ index_url[aid]=url
|
|
17
|
+ for index,score in data["topic_lda"]:
|
|
18
|
+ feature[index]=score
|
|
19
|
+ t.add_item(aid, feature)
|
20
|
20
|
t.build(10)
|
21
|
21
|
t.save('article.ann')
|
22
|
22
|
with open("index_url","w") as f:
|
23
|
|
- pickle.dump(index_url,f)
|
|
23
|
+ pickle.dump(index_url,f)
|
24
|
24
|
|
25
|
25
|
|