基于用户社会关系的社交网络好友推荐算法研究

发布时间：2019-06-07 15:35

【摘要】：社交网络中存在海量用户,如何有效推荐好友是社交网络可持续发展的重要环节,也是社交网络相关研究的重要主题。当前实践及现有研究往往基于用户的显性信息推荐好友,而忽略了用户之间的隐性社会关系;此外,显性信息往往不够完整且存在虚假信息问题。为有效实现好友推荐,本文提出了基于用户社会关系的好友推荐算法,并重点应用关联规则算法分析用户之间的隐含关联度,构造用户之间的网络有向图及关系转移矩阵;然后,结合关系转移矩阵与PageRank算法计算每个用户的分数,将分数较高的用户推荐给目标用户。在此基础上,本文引入用户影响力,提出综合考虑用户社会关系及用户影响力的PeopleRank算法。为验证算法的合理性和有效性,将本文所提出的两种算法与传统的社会过滤算法、PageRank算法进行对比分析。为此,本文抓取了Twitter社交网站上用户数据开展实验分析。实验结果显示本文所提出的算法具有较好的推荐效果,尤其是综合考虑用户社会关系及用户影响力的好友推荐算法在推荐准确率和推荐召回率上都有明显的优势。
[Abstract]:There are a large number of users in social networks. How to effectively recommend friends is an important part of the sustainable development of social networks, and it is also an important topic of social network related research. The current practice and existing research often recommend friends based on the explicit information of the user, but ignore the hidden social relations between the users; in addition, the explicit information is often incomplete and there is a problem of false information. In order to effectively realize friend recommendation, this paper proposes a friend recommendation algorithm based on user social relations, and focuses on the application of association rules algorithm to analyze the implicit correlation degree between users, and construct the network directed graph and relationship transfer matrix between users. Then, the relational transfer matrix and PageRank algorithm are combined to calculate the scores of each user, and the users with higher scores are recommended to the target users. On this basis, this paper introduces user influence, and proposes a PeopleRank algorithm which considers user social relations and user influence synthetically. In order to verify the rationality and effectiveness of the algorithm, the two algorithms proposed in this paper are compared with the traditional social filtering algorithm and PageRank algorithm. For this reason, this paper grabs the user data on Twitter social networking site to carry on the experimental analysis. The experimental results show that the algorithm proposed in this paper has a good recommendation effect, especially the friend recommendation algorithm which takes into account user social relations and user influence has obvious advantages in recommendation accuracy and recommendation recall rate.
【作者单位】：上海大学悉尼工商学院;安徽大学商学院;
【基金】：国家自然科学基金面上资助项目(71371010,71571115) 上海市科学委员会科技人才计划项目(14PJ1403700) 上海市教育委员会科研创新项目(14YS006) 教育部在线教育研究中心在线教育研究基金(全通教育)项目资助(2016YB138)
【分类号】：C912.3

【相似文献】