基于内容的社交网络用户身份识别方法  

Content-Based Social Network User Identification Methods

作　　者：张树森[1] 梁循[1] 弭宝瞳 赵吉超 周小平[1] ZHANG Shu-Sen;LIANG Xun;MI Bao-Tong;ZHAO Ji-Chao;ZHOU Xiao-Ping(School of Information , Renmin University of China , Beijing 100872)

机构地区：[1]中国人民大学信息学院

出　　处：《计算机学报》2019年第8期1739-1754,共16页Chinese Journal of Computers

基　　金：国家自然科学基金(71531012,71271211,71601013);北京市自然科学基金(4172032,4174087);北京教委科技计划项目(SQKM201710016002)资助~~

摘　　要：社交网络中识别用户身份具有重要价值,它对社交网络的分析与监管、用户行为的预测以及用户之间交互过程的研究具有重要意义.该文针对社交网络中的用户身份进行研究,将用户身份分为组织用户和个人用户,并对这两种用户身份进行具体定义和识别.该文研究问题属于社交网络用户分析研究中的子研究问题,主要通过用户在社交网络中发表的文本内容、多媒体内容以及用户时间序列内容识别出该用户的组织-个人身份,为社交网络用户身份的识别及进一步研究提供借鉴和帮助.在识别过程中,通过对文本内容中用户的口语化水平、内容(主题)复杂化水平、内容规范化水平的度量以及多媒体内容中用户图片特性和用户时间序列内容的分析,从不同角度提出5种机器可操作的用户组织-个人身份识别方法,进而识别出社交网络中用户是组织用户还是个人用户.最后,为了验证该文所提识别方法的可行性和有效性,该文选择新浪微博数据进行实验,并通过概率模型识别方法进行了对比分析.同时,在验证过程中,使用多种指标对实验结果进行评价.实验结果表明,该文识别方法能够有效识别出用户的组织-个人身份,其中内容复杂特性识别方法、内容规范化识别方法以及时间序列内容识别方法的用户身份识别准确率超过80%.It is of great value in social networks to recognise user’s identity,which can help analyze and understand social networks,predict user’s behavior,study the supervision of social networks and interaction between users.In this paper,we take user’s identity in social networks as the research object.In view of the text content,multimedia content and time series content user post,we study the user’s identity in social networks,which is divided into organization user and individual user,and the two user identities are defined and identified concretely.The problem in this paper is a sub-research problem in social network user analysis,and it mainly identifies the user’s organization or individual identity through text content,multimedia content and time-series content published by social network users,which provided reference and help for the identification of social network user identity and further study.In social life,everyone has an identity in a specific environment relative to other people or things.Due to the different nature of users,organization users and individual users will have different characteristics in terms of social content.In social networks,user identities are essential parts of their social activities.User identities may be either explicit or implicit,and are more or less expressed in user-published content.At the same time,more user characteristics are also reflected in user postings.Therefore,the identity of social network users can be analyzed and identified through the user’s content.There are many kinds of information that can reflect the user’s identity in social networks,such as tag information,registration information,labels or personal statements.However,it is difficult for us to identify the authenticity or credibility of these information.The registration information and authentication information in different social platforms are also different,so it is difficult to find common information that is suitable for many social platforms. However,the content information publish

关 键 词：社交网络 身份识别 内容复杂化 内容规范化 时间序列 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]