摘要
针对Web用户聚类时,社会标注系统中用户访问资源数据稀疏从而导致传统聚类算法效率不高的问题,提出了一种三向迭代聚类算法,对用户、标签和资源分别聚类,利用三者之间的关联关系不断相互交叉迭代调整,直到各聚类簇达到稳定为止。实验表明,该方法调整后类的内聚性更强,区分度更大,能有效解决数据稀疏性问题,提高用户聚类效果。
Aiming at the problem of the sparse users' access resource data resulting in low efficiency of traditional clustering algorithms,this paper proposed a tripartite iterative clustering algorithm, which cluster tags, resources and users respectively and then used the relations among them to cross iteratively adjust continuously, until all clusters achieved stability status so that the distances within the cluster were much smaller whereas the distances between the clusters were even bigger. Experiment shows that this method can effectively solve the data sparse problem, and improve the effect of user clustering.
出处
《计算机应用研究》
CSCD
北大核心
2013年第12期3557-3559,3592,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(61272277)
安徽省教育厅优秀青年基金重点项目(2011SQRL117ZD)
关键词
社会标注
大众分类
三方网络
迭代聚类
social tagging
folksonomy
tripartite network
iterative cluster