基于协同过滤的Web服务动态社区发现算法

发布时间：2018-09-09 13:24

【摘要】：针对现有社区发现算法挖掘结果精确度不高以及Web服务资源智能推荐质量较低的问题,在传统协同过滤算法的基础上,提出了基于节点相似性的动态社区发现算法。首先以连接节点最多的中心节点为起始网络社区,以社区贡献度为衡量指标不断形成多个全局贡献度饱和的社区;再使用重叠度计算将相似度高的社区进行合并,最后通过计算目标用户与社区中其他用户之间的动态相似度,将计算结果降序排列后构成邻近用户集,获得社区化推荐对象。实验结果表明,提出的社区发现算法对用户社会网络的社区分类与实际社区分类结果吻合,提高了社区挖掘的精确度,有助于实现高质量的社区化推荐。
[Abstract]:Aiming at the problems of low precision of mining results of existing community discovery algorithms and low quality of intelligent recommendation of Web service resources, a dynamic community discovery algorithm based on node similarity is proposed on the basis of traditional collaborative filtering algorithm. Firstly, the central node with the most connected nodes is taken as the starting point of the network community, and the contribution of the community is taken as the balance. Quantitative indicators constantly form multiple communities with saturated global contributions, and then use overlap calculation to merge communities with high similarity. Finally, by calculating the dynamic similarity between the target users and other users in the community, the results are arranged in descending order to form a neighboring user set and the community recommendation objects are obtained. The proposed community discovery algorithm is consistent with the actual community classification results of user social networks, which improves the accuracy of community mining and helps to achieve high-quality community recommendation.
【作者单位】：武汉理工大学经济学院;武汉理工大学华夏学院经济与管理系;
【基金】：国家自然科学基金资助项目(71072077,71172043) 国家科技支撑计划项目(2011BAH16B02) 教育部留学回国人员科研启动基金资助项目(20101561) 中央高校基本科研业务费专项资金资助项目(2012YB20)
【分类号】：TP393.09

【参考文献】