基于非负多矩阵分解的微博网络信息推荐

发布时间：2018-07-26 16:15

【摘要】：微博网站作为一种流行的社交媒体形式,在为用户提供丰富信息和服务的同时,也带来了信息超载问题。如何利用微博网络为用户推荐有价值的信息,以缓解信息超载问题变得日益重要。根据微博网络的有向性以及建立关注关系的随意性等特点,提出了一种基于非负多矩阵分解的微博网络推荐方法,综合考虑了用户之间的关注关系、用户与微博内容的转发关系,以及微博内容与主题的所属关系等多源信息。基于新浪微博数据集进行了微博内容推荐实验,结果表明基于非负多矩阵分解的方法能够有效利用微博网络中的多维信息,显著提高推荐准确度。该方法不仅能挖掘出微博内容的主题,还能挖掘出用户间的关联关系,可推广到对用户进行好友和主题的推荐。
[Abstract]:As a popular form of social media, Weibo website not only provides users with abundant information and services, but also brings the problem of information overload. How to use Weibo network to recommend valuable information for users to alleviate the problem of information overload has become increasingly important. According to the characteristics of Weibo network and the randomness of establishing the relationship of concern, this paper proposes a Weibo network recommendation method based on non-negative multi-matrix decomposition, which considers the relationship of concern among users. The forwarding relationship between user and Weibo content, and the relationship between Weibo content and topic. The experiment of Weibo content recommendation based on Sina Weibo dataset is carried out. The results show that the method based on non-negative multi-matrix decomposition can effectively utilize the multidimensional information in Weibo network and improve the accuracy of recommendation. This method not only can mine the topic of Weibo content, but also can mine the relationship between users, and it can be extended to recommend friends and topics to users.
【作者单位】：云南大学信息学院;云南红岭云网络科技有限公司;
【基金】：云南省科技创新强省计划资助项目(2014AB016) 国家自然科学基金资助项目(61562090) 云南省应用基础研究计划面上项目(2013FB009)
【分类号】：TP391.3;TP393.092

【相似文献】