基于关系图特征的微博水军发现方法

发布时间：2017-12-31 20:26

本文关键词：基于关系图特征的微博水军发现方法　出处：《自动化学报》2015年09期 　论文类型：期刊论文

【摘要】：随着网络水军策略的不断演变,传统的基于用户内容和用户行为的发现方法对新型社交网络水军的识别效果不断下降,水军用户可以变更自身的博文内容与转发行为,但无法改变与网络中正常用户的连结关系,形成的结构图具有一定的稳定性,因此,相对于用户的内容特征与行为特征,用户关系特征在水军识别中具有更强的鲁棒性与准确度.由此,本文提出一种基于用户关系图特征的微博水军账号识别方法.实验中通过爬虫程序抓取新浪微博网络数据;然后,提取用户的属性特征、时间特征、关系图特征;最后,利用三种机器学习算法对用户进行分类预测.仿真结果表明,添加新特征后对水军账号的识别准确率、召回率提高5%以上,从而验证了关系图特征在水军识别中的有效性.
[Abstract]:With the continuous evolution of the strategy of the network navy, the traditional methods based on user content and user behavior to identify the new social network navy are declining. Marine users can change their blog content and forwarding behavior, but can not change the link with the normal users in the network, the resulting structure has a certain stability, so. Compared with the content and behavior characteristics of users, the user relationship features have stronger robustness and accuracy in the identification of water army. In this paper, a method of identifying the Weibo water army account based on the feature of user relationship graph is proposed. In the experiment, the data of Sina Weibo network is captured by crawler program. Then, the attribute feature, time feature and graph feature of the user are extracted. Finally, three kinds of machine learning algorithms are used to classify and predict the users. The simulation results show that the recognition accuracy of the waterfront accounts is improved by more than 5% after adding new features. Thus, the validity of the relationship diagram feature in the recognition of water army is verified.
【作者单位】：国家数字交换系统工程技术研究中心;
【基金】：国家高技术研究发展计划(863计划)(2011AA010604) 国家科技重大专项(2013ZX03006002)资助~~
【分类号】：D669;TP393.092
【正文快照】： 引用格式程晓涛,刘彩霞,刘树新.基于关系图特征的微博水军发现方法动化学报,2015,4 1(9):1 5 3 3-1541随着社交网络的迅猛发展,信息的流动模式与服务模式也发生着重火变化,网比均可借助第三方平台生产信息,使得信息源急剧增大,信息发布方式多样.微博,作为一种备受欢迎的社交网

【参考文献】