Enhancing personalized recommendation in social tagging systems by tag expansion

Recommendation is a technique commonly adopted to address and mitigate the negative effect of information overload problem. Collaborative filtering recommendation technique is one the most successful and popular approaches. However, collaborative filtering approach only considers whether or not a user likes a specific resource but does not take into account the reason why she/he likes it. Social tagging systems, a well-known mechanism in various Web 2.0 applications, play a crucial role to deal with this issue. In response, some prior studies have tried to incorporate social tagging systems into traditional recommendation approaches to enhance their performance. In this study, we investigate the effect of tag completeness in tag-based recommendation approach by proposing a tag-expansion-based personalized recommendation (TE-PR) technique to enrich user profiles and resource profiles. According to our empirical evaluation results using CiteULike data set, the tag-expansion-based approach indeed improves the performance of recommendation techniques. Moreover, the expanded resource profiles have significant positive contributions on recommendation performance.