融合事件关系推理和情感博弈的网络不实信息演化机理研究

项目名称： 融合事件关系推理和情感博弈的网络不实信息演化机理研究

项目编号： No.61502151

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 计算机科学学科

项目作者： 杨玉珍

作者单位： 菏泽学院

项目金额： 20万元

中文摘要： 自媒体环境下，由于信息呈碎片化传播导致网民难以认清事件发生的前因后果，为网络不实信息的滋生和传播提供可乘之机。研究网络不实信息演化机理对探测民意、净化网络空间、维护社会稳定等方面具有积极的现实意义。. 本课题从事件关系推理和情感博弈的角度分析网络不实信息的演化机理，首先针对由信息碎片化传播带来的事实真相获取困难的问题，研究同一事件关联检测机制，提出一种基于实体依赖的层次关联检测模型，并结合情感一致性来推理事件之间的关系；其次针对不实信息事件关系不唯一问题，研究网民历史信息和历史行为，挖掘网民间的强弱关系作为初始博弈角色定义的依据；进一步研究网民个体情感和群体情感，在定义评价对象潜在情感的基础上推理网民隐性情感，构建基于情感博弈的角色转换模型，用于划分博弈角色团体作为事件关系消歧的依据；最后，构建信息信任评估模型，用于预判网络不实信息，还原事实真相。

中文关键词： 情感博弈；事件关系推理；兴趣度分析；角色转移；团体划分

英文摘要： In the media environment, information spreads fragmentarily. It is difficult for netizens to fully get the ins and outs of events, while it is convenient for the network false information to breed and spread. Thus, studing the evolution mechanism of network misinformation has a realistic significance for detecting public opinion, purifying network information space, and maintaining social stability.. This paper analyzes the evolution mechanism of network misinformation from the perspectives of the events relations and emotional game. First, to solve the problem of unavailable event truth brought by the fragmentation spreading of information, we study the correlation detection model on the same event and propose a hierarchy correlation detection model which is entity-dependent, on this basis, combine the emotional consistency to reason the relationship between events. Second, against the problem that the event relationship of false information is not exclusive, we study the history of Internet information and historical behaviors of netizens, excavate the strength of the relationship between Internet users as the basis of role definitions of initial game. Furthermore, we study the individual emotions and group emotions of netizens, infer their potential emotions on the basis of defining the potential emotions of evaluation objects, and build the role transformation model based on emotional game, to divide the game role group and provide an evidence for the disambiguation of event relationship. Finally, we construct the trust evaluation model for information, and use it for predicting network false information and restoring the truth.

英文关键词： emotional game;events relations;interestingness analysis;role transformation;game group division