Kinetic Modeling of an Opinion Model on Social Networks

It is commonly accepted that, on social networks, the opinion of the agents with a higher connectivity, i.e., a larger number of followers, results in more convincing than that of the agents with a lower number of followers. By kinetic modeling approach, a kinetic model of opinion formation on social networks is derived, in which the distribution function depends on both the opinion and the connectivity of the agents. The opinion exchange process is governed by a Sznajd type model with three opinions, ±1, 0, and the social network is represented statistically with connectivity denoting the number of contacts of a given individual. The asymptotic mean opinion of a social network is determined in terms of the initial opinion and the connectivity of the agents.

Two-Layer Coupled Network Model for Topic Derivation in Public Opinion Propagation

In view of the fact that news can generate derivative topics when it spreads through micro-blogs,a two-layer coupled SEIR public opinion propagation model is proposed in this paper.The model divides the process of public opinion propagation into two layers:the original topic layer and the derived topic layer.Messages are transmitted separately by the SEIR model in the two topic layers,which are independent and interactive.The influence of the topic derivation rate on the propagation trend is established by solving for the equilibrium point and propagation threshold.Further,we establish the relationship between the original topic and the derived topic by simulation.This paper uses the Baidu index to demonstrate the correctness of the model.The relationship between the derived topic and the original topic is verified by adjusting the parameters by the control variable method.The results show that the proposed model is consistent with the propagation of actual public opinion.

Direct immune-SCIR public-opinion propagation model based on real-time online users

Current public-opinion propagation research usually focused on closed network topologies without considering the fluctuation of the number of network users or the impact of social factors on propagation. Thus, it remains difficult to accurately describe the public-opinion propagation rules of social networks. In order to study the rules of public opinion spread on dynamic social networks, by analyzing the activity of social-network users and the regulatory role of relevant departments in the spread of public opinion, concepts of additional user and offline rates are introduced, and the direct immune-susceptible, contacted, infected, and refractory (DI-SCIR) public-opinion propagation model based on real-time online users is established. The interventional force of relevant departments, credibility of real information, and time of intervention are considered, and a public-opinion propagation control strategy based on direct immunity is proposed. The equilibrium point and the basic reproduction number of the model are theoretically analyzed to obtain boundary conditions for public-opinion propagation. Simulation results show that the new model can accurately reflect the propagation rules of public opinion. When the basic reproduction number is less than 1, public opinion will eventually disappear in the network. Social factors can significantly influence the time and scope of public opinion spread on social networks. By controlling social factors, relevant departments can analyze the rules of public opinion spread on social networks to suppress the propagate of negative public opinion and provide a powerful tool to ensure security and stability of society.

Ising model on evolution networks and its application on opinion formation

Many phenomena show that in a favorable circumstance an agent still has an updating possibility, and in an unfavor- able circumstance an agent also has a possibility of holding its own state and reselecting its neighbors. To describe this kind of phenomena an Ising model on evolution networks was presented and used for consensus formation and separation of opinion groups in human population. In this model the state-holding probability p and selection-rewiring probability q were introduced. The influence of this mixed dynamics of spin flips and network rewiring on the ordering behavior of the model was investigated, p hinders ordering of opinion networks and q accelerates the dynamical process of networks. Influence of q on the ordering and separating stems from its effect on average path length of networks.

An Overview on Opinion Spreading Model

Research on opinion spreading has received more and more attention in recent years. This paper focus on make a summary of opinion evolution researches, we first review some classical opinion models, and then introduce the existing result of improvement models from the aspect of opinion space, model parameter, social network and so on. The current study’s limitation and further research are also prospected at the end. By in-depth understand the opinion spreading mechanism so as to guide and control the public opinions, which is very useful and meaningful.

Blog Opinion Retrieval with Generation Model and Mixture Model

Blog opinion retrieval aims to find blogs with opinionated information related to a given topic.Its main problem is to compute the opinion score,which balances topic relevance and opinion relevance.To deal with this problem a generative model deduced by a Bayesian approach is pro-posed,and an improved mixture model is proposed to estimate the opinion relevance between a blog and a given topic in our retrieval framework.Moreover,pointwise mutual information is used to expand sentiment words for different topics based on a general sentimental lexicon.The correlation between topic and candidate words is applied in the process of both expanding sentiment words and estimating sentence opinion scores.Experimental results show that the proposed approaches improve upon the state-of-the-art opinion retrieval method on TREC2010 dataset.

基于改进SEIQR模型的突发网络舆情传播与控制研究

[目的/意义]社会突发事件映射到互联网空间中形成的突发网络舆情在不良因素的干扰下于网民间扩散容易形成强烈的负面情绪导向,很可能对网络社会的和谐稳定产生极大破坏。因此,正确认识突发网络舆情的传播特点,把握各种舆情管控措施对传播的影响,对合理、有效地管理舆情至关重要。[方法/过程]针对突发网络舆情的传播特性对标准SEIQR模型进行改进,提出实施舆情管控措施下以媒体恶意炒作和网民间恶性正反馈为主要传播驱动的SEACBR模型,并利用仿真实验对不同规模的突发网络舆情在各种管控措施下的传播过程进行模拟,以探究不同措施的管控效果并验证模型的泛化性。[结果/结论]管控突发网络舆情应以良性引导为主,其他措施为辅,这样在及时施策、积极治理的情况下突发网络舆情的消极影响就可被有效抑制。

考虑用户能动性和流动性的舆情传播模型

针对现有信息传播模型忽略了用户主观能动性和社交网络动态性的问题,提出异构网络中考虑用户能动性和流动性的SCBRD(Susceptible-Commented-Believed-Recovered-Defensed)舆情传播模型。首先,利用下一代矩阵方法计算基本再生数,并运用Lyapunov稳定性定理和庞特里亚金原理分析了系统的动力学和最优控制问题。然后,基于BA(Barabási-Albert)无标度网络进行仿真分析以确定影响舆情传播的显著因素,结果表明,用户的好奇心理、转发行为和进入率在信息扩散中起着主导作用,并且系统存在最优控制解。最后,依据实际数据验证模型的合理性。与SCIR(Susceptible-inCubation-Infective-Refractory)模型相比,SCBRD模型的拟合优度提高了27.40%,预测的均方根误差(RMSE)减小了39.02%。因此,该模型能够适应信息传播复杂的变化形势,为官方的舆情监管提供较好的指导。

基于改进HK模型的社交用户观点演化预测研究

[目的/意义]准确预测舆情观点的演化态势,对化解舆情危机具有重要意义。[方法/过程]针对HK模型用于用户观点演化预测存在的不足,进行如下改进:首先,从和当前事件相似度高于阈值的历史博文中获取用户的初始观点值;其次,利用用户间交互行为以及关注关系计算亲密度,修订用户信任阈值;最后,融合用户全局影响力、用户间交互度以及观点接近度计算用户间的综合信任度,突出用户间观点影响权重的差异性。最终利用改进的HK模型实现对社交用户观点演化预测。[结果/结论]实验结果表明,改进的HK模型用于观点演化预测具有较低的MSE值和MAE值,可为舆情发展提供有效预测。

一种干扰系数自探测的网络事件选取方法

网络事件在传播过程中会产生相互影响,对这种事件间的影响进行研究有助于深入了解网络舆情的传播规律,并依此采取正确的应对办法.针对不同事件间存在不同干扰效果这一问题,提出了一种干扰系数自探测的网络事件选取方法.该方法使用具有多事件容量特征的mSIR-CA(MultipleSIR-Cellular Automata)模型为推演核心,并借助真实事件数据进行有针对性的参数寻优,使得方法能够根据目标事件,从当前同处于扩散过程的多个备选干扰事件中,进行有效事件选取及干扰效果排序.实验结果表明,根据真实事件训练集对方法参数寻优后,通过对已发生事件标记的事件组进行验证,该方法在有效干扰事件的分类判断、干扰效果排序结果方面有较好的准确率.

金融网络舆情演化机制仿真与实证

构建了金融网络舆情观点聚合模型并进行模拟仿真,在此基础上以2020年“美股熔断”为现实案例,检验了金融网络舆情观点聚合过程,刻画了金融网络舆情演化机制。研究发现:第一,在仿真环境下,个体接受度越大,舆情观点聚合速率越快;信任阈值越大,网络中的观点更容易趋于一致;个体间的强关系更能推动金融网络舆情观点聚合。第二,实证案例验证了个体投资者间的意见交互有利于实现观点聚合,但他们基于消极和积极情绪所表达出的观点仍存在两极对立,且个体投资者与主流财经媒体之间的强关系更能加速推动金融网络舆情演化。该研究结果对于完善中国金融网络舆情监管具有启示意义。

重大突发事件下网络舆情传播中社会群体主题演化图谱构建研究

研究重大突发事件下网络舆情传播中社会群体主题演化图谱构建,对重大突发事件的应急和维护社会稳定具有重要作用。文章基于LDA主题模型构建社会群体主题演化图谱,基于主题时序统计和主题频次统计构建社会群体主题热度演化图谱,基于JS散度和DFS算法构建社会群体主题路径演化图谱,并给出重大突发事件下网络舆情传播中社会群体主题演化图谱构建过程模型。研究发现:利用构建的模型可以对群体主题特征、主题热度影响因素及主题最优传播路径进行分析,发现重大突发事件下舆情演进态势和识别敏感舆情话题,从而进行正向的舆情引导,为重大突发事件下网络舆情治理提供新的参考视角。

基于在线品牌社区意见领袖的用户关键需求挖掘

随着社会的快速发展以及技术的不断进步,人们生活节奏不断加快,对产品的需求也在快速发生着变化。在线评论是目前用户需求表达的重要渠道。为克服不加区分挖掘所有评论的传统用户需求挖掘方法耗时过长,难以聚焦用户关键需求的问题,从用户关键性与需求关键性的双关键性角度出发,研发了一种基于在线品牌社区意见领袖的用户关键需求挖掘方法以快速获取用户重要需求,简称KEY-DEMANDS-OL。该方法依据帕累托法则,依托意见领袖评论大数据,采用优化的情感程度及初始改进率,结合KANO模型对用户关键需求进行挖掘。该方法不仅考虑了程度副词的语义信息,提高了情感分析的准确率,而且能够完成意见领袖的生成内容与KANO模型的自动整合,实现用户关键需求的获取与分类。研究结果表明,与挖掘所有评论的传统方法相比,KEY-DEMANDS-OL可以快速获取用户的关键需求,为企业制定产品优化方案提供辅助决策支持。

基于线上-线下超网络模型的舆论演化仿真分析

目前在舆论演化模型及其动力学规则的研究中大多数都没有考虑线下网络,导致其无法真实反映舆论传播过程。为解决上述问题,首先利用线上-线下社交网络映射方法得到线上-线下社交网络拓扑结构;然后构建线上-线下超网络模型,集成舆论事件态势信息形成舆论演化仿真环境;最后,利用传播学原理制定线上-线下超网络模型动力学交互规则,进行舆论动力学仿真。将提出型用于事件实例分析,仿真结果与舆论事件实际发展情况之间的余弦相似度达到0.99,平均差值为1%。实验结果表明,所提模型取得了比其他方法更好的仿真效果,验证了线上-线下超网络模型和舆论动力学交互规则的正确性和有效性。

分数卷积灰色预测模型及其在信息传播中的应用

通过构造累加生成卷积序列的离散卷积运算实现了数据序列的累加生成,并推广到分数阶卷积幂以实现分数阶累加生成。离散序列卷积运算统一了累加生成和累减生成,给定阶数即可生成相应的卷积序列。引入离散卷积运算,可以从信号处理的角度来理解累加生成。累加生成系统可视为一个线性滤波器,累加卷积序列的本质是滤波器的单位冲击响应,卷积运算对应滤波过程。在分数累加卷积序列的基础上构建了分数阶灰色预测模型。最后给出数值实验和信息传播中的实际应用案例验证理论的正确性和方法的有效性。

内地青年网络社群涉台舆情的主题分布与情感特征——基于LDA模型和文本情感计算的大数据分析

综合使用LDA模型、文本情感计算等大数据方法与传统数据处理方法,分析内地主流青年网络社群中的舆情文本后发现,当前青年网络社群中的涉台议题表达可归入“政治”“文娱”“对抗”“生活”“交流”五大主题。在内容特征方面,“政治”和“交流”主题较易获得观看者的支持;“文娱”议题呈现“粉丝向”特征;“对抗”议题较易获得短期关注与转发;“生活”议题较易引发广泛讨论。在情感特征方面,涉台讨论呈现出“爱憎分明”的特点,最常被表达的情感是“赞扬”和“贬责”,但在积极向度的表达中,约有20%—25%属于揶揄和讽刺性用法。在互动形式方面,负面、强烈的情感表达较易引发关注和讨论。另外,用户性别与涉台讨论的情感向度显著相关。未来有必要进一步加强对青年网络社群的涉台舆情追踪,重视舆情引导,并强化青年网络社群中涉台交流的机制建设和积极情感表达。

基于顽固个体的舆论演化动力学研究

由于经典Hegselmann-Krause模型在研究观点演化时没有考虑到个体差异、社会影响力、信息传播的复杂机制等因素,我们在该模型中引入一定比例的顽固个体,并通过网络结构来刻画个体作用之间的差别,进而研究异质群体的舆论演化动力学行为,并分析网络结构和顽固个体比例对舆论演化动力学的影响。结果表明,在没有顽固个体的情况下,高信任阈值会促进共识态的形成,而顽固个体的存在则会导致群体观点的碎片化分布,且高信任阈值并不有利于群体形成局部共识。此外,对比基于度异质性引入顽固个体的不同选择策略下的研究结果发现,选择度较大的个体作为顽固个体时最有利于群体形成局部共识,网络结构、顽固个体比例和顽固个体的选择策略对系统弛豫时间也有显著影响。这些发现对于揭示真实社会系统中舆论演化的宏观规律和内在机理具有重要意义。

基于BERTopic模型的网络暴力事件衍生舆情探测

[研究目的]在海量用户生成内容中及时探测和剖析网络暴力事件的衍生舆情能够为舆情事件链的演化分析、同类舆情的研判介入、衍生事件的监测预警提供理论支持。[研究方法]使用BERTopic模型对短文本内容主题建模并采用聚类的方式展示主题的潜在层次结构。根据词向量余弦相似度设计主题衍生度的计量算法,同时融合词共现网络在文档-词语层面信息捕捉的优势以及桑基图直观演示舆情演化过程的特点,衡量主题间的影响力与衍生关系。[研究结论]在开源数据集下多组主题模型的对照实验中,BERTopic模型在短文本建模以及下游任务的平均得分提高2.13%。在网络暴力热点事件的应用实例中,多维细粒度分析与交互式可视化方法可达到直观展示暴力事件的主题聚类、词义关联与演化态势的效果,实现网络暴力事件衍生舆情的探测与分析。

基于知识蒸馏模型的文本情感分析

为了解决预训练语言模型训练时间过长、参数多且难以部署,以及非预训练语言模型分类效果较差的问题,提出了基于知识蒸馏模型的文本情感分析。以预训练深度学习模型(Bidirectional Encoder Representations from Transformers, BERT)作为教师模型,选择双向长短期记忆网络(Bidirectional Long Short-Term Memory, BiLSTM)作为学生模型;在知识蒸馏过程中,将教师模型的Softmax层的输出作为“知识”蒸馏给学生模型,并将蒸馏后的模型应用到公共事件网络舆情文本情感分析中。实验结果表明,该模型参数仅为BERT模型的1/13,使BiLSTM模型的准确率提升了2.2百分点,优于其他同类别轻量级模型,提高了文本情感分析效率。