Information discovery based on Social Bookmarks

The quality of Web was changed by recent social services such as blog, wiki, SNS, and Social bookmark. The user was shifting to creator of the information resource from onlooker. The amount of the information resources on the Web are rapidly increasing day by day. Social services accumulate folk's trends or individual preferences, then social bookmark data can use be an important data mining resource, and it may be possible to discover valid data for recommendation, marketing, and trend analysis . The authors propose two methods for information discovery from a social bookmark. The first method discovers similar users, and also discovers prefer pages which are preferred by a user. In this method, one user's bookmarks are considered as the profile of the user, and calculate similarity between users using profiles. The authors also propose page recommendation using user similarity. The second method recommends newly arrived information. This method discovers the alpha-bookmarkers who are an early detector of popular pages and also a spreader of the pages. This method considers not only earliness but also a hierarchy of interests. Especially, this method calculates neighborhoodness between a user's and an alpha-bookmarker in the specific interest. The authors implemented the two methods and examine using real data retrieved from "HATENA Bookmark". Examination shows that proposed methods are effective. Keyword Social bookmarking, Folksonomy, Information discovery, Alpha-bookmarker