Tag recommendation based on social comment network

Tagging has rapidly become a popular way to annotate the content on social network sites. Tags describe the contents of the resource or provide additional contextual and semantical information to make the content more easily browsable and discoverable by others. However, as tagging is not constrained by a controlled vocabulary and a certain number of resources have no comments, tags tend to be noisy and sparse. In this paper we show that a substantial level of local lexical and topical alignment is observable among users who lie close to each other in the social comment network. We analyze a representative snapshot of Flickr and construct comment networks that present the user clusters in which users add comments to the central user's photos. We find similar users with common interests based on K-Nearest Neighbor and present tag recommendation strategy in local comment context. The results of the evaluation show that tag recommendation based on local lexicon is an effective way to improve collaborative resource sharing in social network.