基于把关人行为的微博虚假信息及早检测方法

Misinformation Detection Based on Gatekeepers' Behaviors in Microblog

目前微博已成为人们获取信息和发布信息的一个重要平台,然而微博也正成为虚假信息滋生和泛滥的温床.现有的方法主要基于分类算法来识别虚假信息,这些方法不能及早发现微博上流行的虚假信息.为了减少虚假信息对公众的影响,使微博在人们的生产和生活中发挥更积极的作用,文中提出一种基于把关人行为的微博虚假信息及早检测方法.该方法利用模型状态持续时间概率为Gamma分布的隐半马尔可夫模型来刻画信息转发者和评论者对流行的真实信息的把关行为,基于此来及早识别微博上流行的虚假信息.该方法分为模型训练和虚假信息检测两个阶段,在虚假信息检测阶段,计算每条信息在传播过程中产生的观测序列相对于模型的平均对数似然概率,实时更新每条信息的可信度,从而及早发现虚假信息,降低虚假信息的危害.使用采集的新浪微博数据集和Twitter数据集对文中的方法进行了测试,实验结果表明了该方法的有效性.

Nowadays,microblog has become a popular social medium for information dissemination.However,it is also one of the main ways of misinformation transmission,due to the lack of an effective scheme for detection and countermeasures.Existing methods mainly detect misinformation based on classification algorithms.They can＇t identify the popular misinformation early.In order to reduce the harmfulness of misinformation and make microblog more user-friendly,in this paper a new method is presented for detecting misinformation based on gatekeepers＇ behaviors.In the proposed scheme,hidden semi-Markov model is used to describe the behaviors of the forwarders and reviewers of popularly true information.Gamma distribution is introduced to describe the state duration of the model.The proposed method includes a training phase and a detection phase.In the detection phase,the average log likelihood of every observation sequence is calculated,and the credibility value of information is updated in real time.So this method can identify the popular misinformation early and reduce the harmfulness of misinformation.An experiment based on real datasets of Sina Weibo and Twitter is conducted to evaluate this method.The experiment results validate the effectiveness of this method.