Identifying user interests from online social networks by using semantic clusters generated from linked data

Abstract

Recently, online social network services (SNSs) are being spotlighted as a means to understand users' implicit interests out of abundant online social information. Since SNS contents such as message posts and comments are however less informative comparing with news articles and blog posts, it is difficult to identify users' implicit interests by analyzing the topics of the SNS contents of users. In this paper, we propose a semantic cluster based method of combining SNS contents with Linked Data. By traversing and merging relevant concepts, the proposed method expands keywords that are helpful to understand the topic similarity between SNS contents. By using Facebook data, we demonstrate that the proposed method increases the coverage of potential interests by 28.85% and the user satisfaction by 17.24% compared to existing works. © Springer International Publishing 2013.