基于H因子的微博社区发现方法
发布时间：2019-03-22 13:33
【摘要】：近年来,微博己逐渐成为社交网络的核心。其从传统的社交网络中脱颖而出,在拥有了独立的服务平台后逐渐演化为一种新的信息发布形式。目前中国微博的注册用户数量已突破5亿,其平台中存在大量有价值的信息可以发掘。 在微博网络在形成过程中,部分微博用户会逐步形成一种小团体结构。微博网络中的小团体结构是社会网络中的一种的社区现象,如果能够挖掘到具有相同或相似兴趣爱好的小团体,就能更好的帮助微博用户选择关注对象,同时可以对具有相同兴趣爱好的用户群体进行精准的广告投放,方便微博营销工作的开展。 为了满足婴幼儿产品微博营销寻找投放目标的需求,本文提出了一种基于用户影响力的微博社区发现方法。本文通过一种基于用户影响力的主题社区发现方法,利用本文构建的基于用户行为的僵尸粉识别模型,剔除了社区中的僵尸粉用户,保证了社区的纯洁性。本文提出的社区发现算法结合了基于H指数传播能力的用户影响力排名、基于支持向量机的文本分类器、基于用户行为的僵尸粉识别模型等算法,经过真实数据的采集,最后从不同的维度对结果数据进行了实验及分析,通过实验分析表明基于本文提出的用于影响力的社区发现方法,具有较高的效率。
[Abstract]:In recent years, Weibo has gradually become the core of social networks. It stands out from the traditional social network and gradually evolves into a new form of information release after having an independent service platform. At present, the number of registered users of Weibo in China has exceeded 500 million, and there are a lot of valuable information in its platform. In the formation of Weibo network, part of Weibo users will gradually form a small group structure. The structure of small groups in Weibo's network is a community phenomenon in the social network. If we can dig up small groups with the same or similar interests, we can better help Weibo users choose the objects of concern. At the same time, the user group with the same interests can carry out accurate advertising to facilitate the development of Weibo's marketing work. In order to meet the demand of infant product Weibo marketing to find the target, this paper presents a method of Weibo community discovery based on user influence. In this paper, by using a subject community discovery method based on user influence and the model of corpses powder recognition based on user behavior constructed in this paper, the users of zombie powder in the community are eliminated, and the purity of the community is guaranteed. The community discovery algorithm proposed in this paper combines the ranking of users' influence based on H-index propagation ability, the text classifier based on support vector machine (SVM) and the recognition model of zombie powder based on user's behavior. Finally, the result data are tested and analyzed from different dimensions. The experimental analysis shows that the proposed community discovery method based on this paper has a high efficiency.
【学位授予单位】：东北林业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092
本文编号：2445632
[Abstract]:In recent years, Weibo has gradually become the core of social networks. It stands out from the traditional social network and gradually evolves into a new form of information release after having an independent service platform. At present, the number of registered users of Weibo in China has exceeded 500 million, and there are a lot of valuable information in its platform. In the formation of Weibo network, part of Weibo users will gradually form a small group structure. The structure of small groups in Weibo's network is a community phenomenon in the social network. If we can dig up small groups with the same or similar interests, we can better help Weibo users choose the objects of concern. At the same time, the user group with the same interests can carry out accurate advertising to facilitate the development of Weibo's marketing work. In order to meet the demand of infant product Weibo marketing to find the target, this paper presents a method of Weibo community discovery based on user influence. In this paper, by using a subject community discovery method based on user influence and the model of corpses powder recognition based on user behavior constructed in this paper, the users of zombie powder in the community are eliminated, and the purity of the community is guaranteed. The community discovery algorithm proposed in this paper combines the ranking of users' influence based on H-index propagation ability, the text classifier based on support vector machine (SVM) and the recognition model of zombie powder based on user's behavior. Finally, the result data are tested and analyzed from different dimensions. The experimental analysis shows that the proposed community discovery method based on this paper has a high efficiency.
【学位授予单位】：东北林业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092
【参考文献】
相关期刊论文 前10条
1 张晨逸;孙建伶;丁轶群;;基于MB-LDA模型的微博主题挖掘[J];计算机研究与发展;2011年10期
2 许晓东;肖银涛;朱士瑞;;微博社区的谣言传播仿真研究[J];计算机工程;2011年10期
3 袁鹏亮;;基于Pagerank网页排名技术计算方法的探讨[J];农业网络信息;2007年12期
4 张金增;范明;;一种改进的基于最大流的Web社区挖掘算法[J];计算机应用;2009年01期
5 王慧芳;黄林鹏;俞晟;;一种增量式的社区发现算法研究[J];计算机仿真;2008年01期
6 皮靖;邵雄凯;肖雅夫;;基于朴素贝叶斯算法的主题爬虫的研究[J];计算机与数字工程;2012年06期
7 高琰;谷士文;唐t;;基于链接分析的Web社区发现技术的研究[J];计算机应用研究;2006年07期
8 马瑞新;邓贵仕;王晓;;基于扩散理论的HITS算法在Web挖掘中的研究与优化[J];计算机应用研究;2012年01期
9 苗家;马军;陈竹敏;;一种基于HITS算法的Blog文摘方法[J];中文信息学报;2011年01期
10 廖福生;江昀;;微博客的信息传播模式及其发展分析[J];宁波广播电视大学学报;2010年02期
，本文编号：2445632
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2445632.html
