Temporal Top-k Search in Social Tagging Sites Using Multiple Social Networks

In social tagging sites, users are provided easy ways to create social networks, to post and share items like bookmarks, videos, photos and articles, along with comments and tags. In this paper, we present a study of top-k search in social tagging sites by utilizing multiple social networks and temporal information. In particular, besides the global connection, we consider two main social networks, namely the friendship and the common interest networks in our scoring functions. Based on the degree of participation in various networks, users can be categorized into specific classes that differ in their weights on each scoring component. Temporal information, usually ignored by previous works, can enhance the popularity and freshness of the ranking results. Experiments and evaluations on real social tagging datasets show that our framework works well in practice and give useful and intuitive results.