随着YouTube、Flickr和Last.fm等社会化网络的兴起,标签系统在日常生活中扮演着越来越重要的作用.为了给用户提供更优质的推荐,分析用户为不同资源打标签的行为就显得尤为重要.本文将主要的社区发现算法应用到标签系统中的聚类分析中,并比较它们在不同数据集上的表现,设计出针对标签系统的个性化推荐算法.实验结果表明,本文提出的算法能很好的发现不同用户的兴趣,提高推荐系统的质量.
With the rise of YouTube, Flickr, Last.fro and other social networks, tagging systems play an increasingly important role in our everyday life. Analyzing user's tagging behavior of different resources is very important in providing high quality services. In this paper, major community structure detection algorithms are implemented in clustering analysis in tagging system. By comparing their performances on different datasets, a personalized recommendation algorithm for tagging system is designed. Experimental results indicate that the proposed algorithm performs well in detecting different user interests and thus enhances the quality of the recommendation system..