基于二分网络集团化的推荐算法改进研究
首发时间:2016-04-07
摘要:信息超载是目前互联网用户面对的一个非常严重的问题,推荐系统的出现使这一问题得到了有效的解决,同时受到了众多学者和工程师的广泛关注。在推荐系统领域中,能够提高推荐算法的精确性与多样性的个性化推荐系统更是成为了大家研究的热点。本文基于同一个集团内的节点更具相似性这一思想,对推荐算法的推荐结果进行了优化,提出了一种基于网络集团化的改进个性化推荐算法。与已有的不同推荐算法在人工网络与实际网络上进行比较,其最高可以提高20%的精确度与7%的多样性,同时揭示了在网络中进行推荐时,根据节点的不同属性进行分类的重要性。
For information in English, please click here
Research on the improvement of recommendation algorithms based on the communities of bipartite networks
Abstract:Information overload is one of the most significant problems for Internet users, the recommender system offers a powerful tool to make this problem well solved and gains wide concerns of scholars and engineers. Personalized recommender system which can improve the accuracy and diversity of recommender algorithms gets more attention. In this article, based on the idea of the nodes within a community share more similarity, we revised the results recommended by the algorithms and proposed a modified personalized recommender algorithm based on the communities of networks. We found that the modified algorithm whether on the artificial benchmarks or real networks performs effectively. The results improve maximum 20% accuracy and 7% diversity and reveal that it is necessary to classify the nodes based on the inherent properties in recommender systems.
Keywords: Complex network Recommender system Bipartite network Community structure
论文图表:
引用
No.4682015114006214****
同行评议
共计0人参与
勘误表
基于二分网络集团化的推荐算法改进研究
评论
全部评论0/1000