earticle

영어

With the rapid development in Web 2.0 application services, tags are used as useful metadata in social collaborative annotation systems. In this paper, we propose a collaborative approach for expanding tag neighbours and investigate the use of a spectral clustering algorithm to deal with issues of ambiguity and redundancy of tags. The algorithm could effectively filter out noisy tag neighbours, and thus obtain more appropriate recommendations for users. Experiments have been conducted on two real world datasets: “MovieLens” and “M-Eco system”, to compare our proposed approach with the traditional collaborative filtering approaches and naive tag neighbours expansion approaches in terms of precision. The results showed the effectiveness and superiority of our proposed method against the traditional approaches, which demonstrate that our approach could considerably improve the performance of recommendation.