Incorporating Sentiment Analysis for Improved Tag-Based Recommendation

Social tagging systems have become as a popular application with the advance of Web 2.0 technologies. By tagging, users annotate and index the resources freely and subjectively, based on their senses of interests, which can improve the performance of the current personalized recommendation systems. In this paper, we propose a sentiment enhanced tag-based recommendation approach by incorporating sentiment analysis of tags that annotated on resources. The presented approach introduces a sentiment enhancement factor to the similarity metric which measures the matching between resources. The evaluation results on a real datasets have demonstrated that our approach can outperform the other compared approaches in terms of recommendation precision.