Boosting tag-based search in social media sites

The availability and pervasive use of smart mobile devices makes it easy to upload videos and photos to the social websites and label them with any arbitrary tags from anywhere and anytime. This paper exploits the social tagging information and reveals the latent hidden tags which might be relevant to a social media item to improve the tag-based search process. The proposed approach predicts links in undirected weighted tripartite graph. From a graph-based proximity perspective, our approach finds the appropriate personalized item in response to the user´s query as well as uncovers the hidden tags potentially relevant to a given item. We evaluate our method on real-world social tagging system collected from MovieLens. The experimental evaluation shows that enriching the low annotated items with hidden tags improves the tag-based search performance.