社交网络中社区领袖的挖掘算法研究

发布时间：2018-06-27 01:29

本文选题：社交网络 + 社区领袖　；参考：《上海交通大学》2014年硕士论文

【摘要】：在社交网络中，用户A关注了用户B，他们之间便产生了联系，众多的联系就会形成社区，社区领袖的挖掘是社交网络的一个研究课题。挖掘社区领袖意味着要识别网络中的重要节点，这就涉及到社区中心性分析，其常用的标准有特征向量中心性。在社交网络中，用户和用户的联系通过关注产生，而在网络上，网页和网页的联系通过链接产生，两者之间有共同性。PageRank算法正是众所周知的用于网页排名的算法，本文为此将其借鉴过来并加以改进生成UserRank算法，使之适用于社区领袖的挖掘。新算法将传统上把影响力平均分配给关注的人的做法，改进为依据用户间的亲密程度不同将影响力按不同的比例分配给关注的人，，从而实现了在用户的关注关系上赋予权重的目的。在社区中，一个用户就是一个节点，节点之间的影响力会互相传递，节点X关注了节点Y，则节点X的影响力就会全部或者部分贡献给节点Y。经过算法多次迭代计算后，社区中每个用户的影响力收敛后趋于稳定，影响力排名最大的用户，就是社区领袖。实验结果表明改进后的新算法能更快更有效地挖掘出社交网络中的社区领袖。
[Abstract]:In the social network, user A pays attention to user B, they have the connection between them, many connections will form the community, the mining of community leader is a research topic of social network. Mining community leaders means identifying important nodes in the network, which involves community-centric analysis. The commonly used criteria are eigenvector centrality. In social networks, the connection between the user and the user is generated by attention, and on the network, the connection between the web page and the web page is generated by the link, and there is a commonality between the two. PageRank algorithm is known as the algorithm used to rank web pages. In this paper, we use it for reference and improve the generated UserRank algorithm to make it suitable for the mining of community leaders. The new algorithm, which traditionally allocates influence equally to people of concern, is improved by distributing influence to people of concern in different proportions, depending on the degree of intimacy between users. Thus, the purpose of giving weight to the user's concern relationship is realized. In the community, a user is a node, the influence between nodes will be transferred to each other, node X pays attention to node Y. then the influence of node X will contribute to node Y. in whole or in part. After iterative calculation, the influence of each user in the community converges and tends to be stable, and the most influential user is the community leader. Experimental results show that the improved algorithm can find community leaders in social networks more quickly and effectively.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【参考文献】