网络社区中的意见领袖特征分析  

A Feature Analysis of the Opinion Leader in On-Line Communities

作　　者：肖宇[1] 许炜[1] 夏霖[1] 

机构地区：[1]华中科技大学电子与信息工程系,湖北武汉430074

出　　处：《计算机工程与科学》2011年第1期150-156,共7页Computer Engineering & Science

基　　金："十一五"科技支撑计划重点项目(2006BAK11B00)

摘　　要：本文通过社会网络分析方法识别网络社区中的意见领袖。首先对意见领袖存在的人际关系网络结构特征进行分析,对比论坛、博客和问答网络之间的区别,提出基于无向、有权重网络模型更能真实准确地识别意见领袖。并基于该网络模型研究和分析了网络论坛结构特征,通过测量其小世界和无标度的复杂网络特征,定量分析意见领袖存在的社会性根源。其次提出了基于无向、有权重网络下的PageR-ank算法,并对比前人提出多种意见领袖识别算法,以某论坛四年历史数据实证了算法的有效性。最后对识别结果进行深入分析,并研究了意见领袖同活跃版块之间的关系,发现通过覆盖少量版块即可覆盖绝大部分意见领袖。We use a method of social network analysis to indentify opinion leaders in social networks. First of all we analyze the characteristics of the social networks of online communities, consider the difference between BBS, blogosphere and Q＆A network, and propose that undirected weighted network models will enhance the identification accuracy of opinion leaders. Then, we analyze the features of social networks. We validate that BBS is smallworld and scalefree, and we make a quantitative analysis of the social background of opinion leaders. We also propose a PageRank algorithm based on undirected weighted networks, and test several algorithms proposed before our experiment. We use four years’ data from a BBS, and test these algorithms. Finally, we make？ a profound analysis and find the relationship between opinion leaders and boards, and the result shows that we can find a majority of opinion leaders from covering few boards.

关 键 词：复杂网络 网络社区 意见领袖 PAGERANK 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]