人肉搜索现象分析建模及组织协作机制研究
发布时间：2018-09-01 06:29
【摘要】：人肉搜索是一个源自中国的基于互联网的社会现象,起源于本世纪初的猫扑论坛,随后迅速发迹,成为一种爆发现象,震撼登场。期间,很多事件引发并导致了一些无意误伤、恶意中伤的恶劣事件,也由此引发了关于是否立法禁止人肉搜索的大规模讨论,但随后人肉搜索在灾后寻亲、网络反腐、舆论监督等领域也大显身手,并很快就在网民自发的约束下走向了正面、公益等健康的发展轨道。人肉搜索是一个典型的群体协作现象,从戏谑娱乐到道德申讨、公共监督,网络互助,人肉搜索经历了波澜壮阔、亦正亦邪的发展并回归到积极多元搜索场景的过程。这些搜索案例为研究大规模群体协作提供了宝贵的素材。本文即以人肉搜索这一大规模群体协作现象作为研究对象,基于近年发生在国内的万余起人肉搜索事件,对参与过人肉搜索的近300万网民形成的动态网民群体组织和参与者之间的协作搜索过程进行了实证与理论研究。针对参与人肉搜索的网民群体,本文研究了人肉搜索网络群体的基本特征,分析了这一群体的结构、角色分工,构建了参与者之间的社交网络,分析了群体的社交关系网络。通过数据实证分析,我们发现群体中绝大多数成员仅参与过几起的事件,不同角色有不同的属性特征,分工也有很大差别,事件发布者媒体属性较强,而参与者则具有很明显的草根性。但这一群体中存在一个较为稳定的成员持续参与并相互配合的核心活跃组织,这一组织成员约占参与者总体的0.3%。核心活跃组织成员之间联系非常紧密,但却不存在权威的从属关系,没有超级领导者,是一个联系紧密但分布性极强的组织。基于这一群体的结构分析,本文进而研究了群体成员在搜索过程中的组织协作机制。通过对人肉搜索事件的传播过程和协作过程的分析,发现每起人肉搜索事件背后的协作者并非一个随机的组织,事实上,人肉搜索的协作过程主要体现为参与者对信息的传播和汇总过程,受传播机制和内容机制双重作用,协作者在协作过程的作用和地位首先与该协作者在参与者之间社交网络中的位置相关,其次与其在协作过程中的信息贡献相关。组织中各协作者的地位有很大差别,其协作网络的特点表明,该协作过程是一个高度自组织的过程,其成长也呈现出多阶段增长等特点。总体而言,人肉搜索协作是一个由个体发起,个人、媒体、组织等多方协作,线下调查,线上汇总求证,寻找最终目标的分布式协作过程。具有线上线下交互频繁、自组织性强等特点。基于实证分析结果,本文对人肉搜索的传播协作过程进行了人工社会建模,基于资源动员模型,建立了人肉搜索相关主体模型和组织模型,在此基础上构建了基于双重偏好机制的协作模型,设计了人肉搜索组织传播动员过程生成算法,利用真实数据测试了该模型,数据显示该模型能够较好地模拟人肉搜索参与者之间协作关系网络及发展过程。经过以上对人肉搜索的实证与模型分析,本文进一步分析了影响人肉搜索结果的相关因素,并基于这些因素训练机器学习模型对人肉搜索结果进行预测,结果表明人肉搜索结果在一定程度上是可以被预测的。这些研究结果揭开了大规模网民协作和动员的神秘面纱,为接下来的研究,尤其是群体智能和网络社会运动相关研究,提供了切入点并提供了一些有意义的发展方向。我们希望更多基于真实数据的相关研究可以得到学界的关注,尤其是来自于社会学,人类学,网络科学,计算机科学,统计物理,控制理论等专业的研究者。多学科的交叉合作将极大的促进相关研究,让人们更好地理解群体协作这个复杂系统,促进人类大规模协作创造出更为辉煌的文明。
[Abstract]:Human flesh search is a social phenomenon originating from the Internet in China. It originated from the cat-poke forum in the early 20th century and then quickly developed into an outbreak and shocked debut. During this period, many incidents triggered and led to some bad incidents of unintentional injury and malicious slander, which also led to whether legislation prohibited human flesh search. After a large-scale discussion, however, human flesh search also became very skilled in the fields of post-disaster relatives seeking, anti-corruption on the Internet, public opinion supervision and so on. It soon became a positive, public welfare and other healthy development track under the restriction of the Internet users'spontaneity. Human flesh search has undergone a magnificent, positive and evil development and returned to a positive multi-search scenario. These search cases provide valuable material for the study of large-scale group collaboration. In this paper, we empirically and theoretically study the collaborative search process between a dynamic group of netizens and participants who participated in human flesh search. Aiming at the netizens who participated in human flesh search, we studied the basic characteristics of human flesh search network group, analyzed the structure, role division and structure of this group. Through data analysis, we find that most of the members of the group have only participated in a few events, different roles have different attributes and characteristics, and the division of labor is also very different. The media attributes of the event publishers are strong, while the participants have very obvious grass. Roots. However, there is a core active organization in which stable members continue to participate and cooperate with each other, accounting for about 0.3% of the total participants. The members of the core active organization are closely linked, but there is no authoritative affiliation, no super leader, and they are closely linked but highly distributed. Strong organization. Based on the structure analysis of this group, this paper further studies the organization and cooperation mechanism of group members in the search process. Through the analysis of the propagation and cooperation process of human flesh search events, it is found that the collaborator behind each human flesh search event is not a random organization. In fact, human flesh search collaboration has occurred. The role and status of collaborators in the collaboration process are first related to the position of the collaborators in the social network among the participants, and secondly to their information contribution in the collaboration process. In general, human flesh search collaboration is a distributed collaboration initiated by individuals, individuals, media, organizations, and other multi-party collaboration, offline investigation, online summary and verification, to find the ultimate goal. Based on the results of empirical analysis, the dissemination and collaboration process of human flesh search is modeled by artificial society. Based on the resource mobilization model, the related agent model and organization model of human flesh search are established. On this basis, the collaboration based on double preference mechanism is constructed. The model is designed to generate the propagation and mobilization process of human flesh search organization. The model is tested with real data. The data show that the model can better simulate the cooperative relationship network and development process between human flesh search participants. Result-related factors, and based on these factors training machine learning model to predict human flesh search results, the results show that human flesh search results can be predicted to a certain extent. We hope that more research based on real data will be of interest to the academic community, especially from sociology, anthropology, network science, computer science, statistical physics, control theory and other professional researchers. It will greatly promote the relevant research, so that people can better understand the complex system of group cooperation, and promote large-scale human cooperation to create a more brilliant civilization.
【学位授予单位】：国防科学技术大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TP391.3;TP393.09
，
本文编号：2216391
[Abstract]:Human flesh search is a social phenomenon originating from the Internet in China. It originated from the cat-poke forum in the early 20th century and then quickly developed into an outbreak and shocked debut. During this period, many incidents triggered and led to some bad incidents of unintentional injury and malicious slander, which also led to whether legislation prohibited human flesh search. After a large-scale discussion, however, human flesh search also became very skilled in the fields of post-disaster relatives seeking, anti-corruption on the Internet, public opinion supervision and so on. It soon became a positive, public welfare and other healthy development track under the restriction of the Internet users'spontaneity. Human flesh search has undergone a magnificent, positive and evil development and returned to a positive multi-search scenario. These search cases provide valuable material for the study of large-scale group collaboration. In this paper, we empirically and theoretically study the collaborative search process between a dynamic group of netizens and participants who participated in human flesh search. Aiming at the netizens who participated in human flesh search, we studied the basic characteristics of human flesh search network group, analyzed the structure, role division and structure of this group. Through data analysis, we find that most of the members of the group have only participated in a few events, different roles have different attributes and characteristics, and the division of labor is also very different. The media attributes of the event publishers are strong, while the participants have very obvious grass. Roots. However, there is a core active organization in which stable members continue to participate and cooperate with each other, accounting for about 0.3% of the total participants. The members of the core active organization are closely linked, but there is no authoritative affiliation, no super leader, and they are closely linked but highly distributed. Strong organization. Based on the structure analysis of this group, this paper further studies the organization and cooperation mechanism of group members in the search process. Through the analysis of the propagation and cooperation process of human flesh search events, it is found that the collaborator behind each human flesh search event is not a random organization. In fact, human flesh search collaboration has occurred. The role and status of collaborators in the collaboration process are first related to the position of the collaborators in the social network among the participants, and secondly to their information contribution in the collaboration process. In general, human flesh search collaboration is a distributed collaboration initiated by individuals, individuals, media, organizations, and other multi-party collaboration, offline investigation, online summary and verification, to find the ultimate goal. Based on the results of empirical analysis, the dissemination and collaboration process of human flesh search is modeled by artificial society. Based on the resource mobilization model, the related agent model and organization model of human flesh search are established. On this basis, the collaboration based on double preference mechanism is constructed. The model is designed to generate the propagation and mobilization process of human flesh search organization. The model is tested with real data. The data show that the model can better simulate the cooperative relationship network and development process between human flesh search participants. Result-related factors, and based on these factors training machine learning model to predict human flesh search results, the results show that human flesh search results can be predicted to a certain extent. We hope that more research based on real data will be of interest to the academic community, especially from sociology, anthropology, network science, computer science, statistical physics, control theory and other professional researchers. It will greatly promote the relevant research, so that people can better understand the complex system of group cooperation, and promote large-scale human cooperation to create a more brilliant civilization.
【学位授予单位】：国防科学技术大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TP391.3;TP393.09
，
本文编号：2216391
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2216391.html
