I downloaded many users' broadcast feed in Google Reader. In a user feed, we can get what articles this user shared and in every article, there is a list of users who like this user. Therefore, we can use a bread-first-search crawl to crawl broadcast feed of all users who have show their preference in more than 1 articles.
I only crawl feeds of users who like Chinese article. After 1 week crawling, I only crawl down 12586 such users and 51690 Chinese articles.
I extract data about what user like what article, and the results show a user like 10 articles on average and a user share 8 articles on average.
Following are some results
users number : 12586
items number : 51690
like records number : 127616
share records number : 99937
没有评论:
发表评论