基于多源信息相似度的微博用户推荐算法

发布时间：2019-07-10 15:48

【摘要】：针对传统的协同过滤(CF)推荐算法中存在的数据稀疏性和推荐准确率不高的问题,提出了基于多源信息相似度的微博用户推荐算法(MISUR)。首先,根据微博用户的标签信息运用K最近邻(KNN)算法对用户进行分类;然后,对得到的每个类中的用户分别计算其多源信息(微博内容、交互关系和社交信息)的相似度;其次,引入时间权重和丰富度权重计算多源信息的总相似度,并根据其大小进行TOP-N用户推荐;最后,在并行计算框架Spark上进行实验。实验结果表明,MISUR算法与CF算法和基于多社交行为的微博好友推荐算法(MBFR)相比,在准确率、召回率和效率方面都有较大幅度的提升,说明了MISUR算法的有效性。
文内图片：

图片说明：多源信息总相似度框架
[Abstract]:In order to solve the problems of data sparsity and recommendation accuracy in traditional collaborative filtering (CF) recommendation algorithm, a Weibo user recommendation algorithm (MISUR). Based on multi-source information similarity is proposed. Firstly, the K nearest neighbor (KNN) algorithm is used to classify the users according to the label information of Weibo users. Then, the similarity of their multi-source information (Weibo content, interaction and social information) is calculated for each class of users. Secondly, the time weight and richness weight are introduced to calculate the total similarity of multi-source information, and the TOP-N user recommendation is carried out according to its size. Finally, the experiment is carried out on the parallel computing framework Spark. The experimental results show that compared with CF algorithm and Weibo friend recommendation algorithm (MBFR) based on multi-social behavior, MISUR algorithm has a great improvement in accuracy, recall rate and efficiency, which shows the effectiveness of MISUR algorithm.
【作者单位】：曲阜师范大学信息科学与工程学院;曲阜师范大学软件学院;
【基金】：国家自然科学基金资助项目(61402258) 山东省本科高校教学改革研究项目(2015M102) 校级教学改革研究项目(jg05021)~~
【分类号】：TP391.3

【相似文献】