社会网络中基于代理的信息传播建模与实验分析

发布时间：2018-06-20 08:21

本文选题：在线社会网络 + 代理建模　；参考：《南京邮电大学》2017年硕士论文

【摘要】：研究信息在社会网络中传播不仅具有理论意义,而且具有应用价值。然而,链路关系的复杂性、传播信息的多样性、个体的异质性都给这项工作的展开带来了一定的困难。近年来,计算机技术的快速发展为复杂网络的研究提供了新的思路,为揭示信息传播规律和预测用户行为提供了可能。本文在建立网络结构模型的基础上,应用基于代理的建模方法模拟信息传播过程,总结传播规律,预测用户行为。首先,文章从社会网络中信息传播的应用场景出发,提炼信息传播与用户决策的一般规律。其次,本文采用链路向量模型对社会网络进行建模。在线社会网络是信息传播的载体,构建贴近现实的社会网络结构有助于再现更加真实的传播规律。本文将节点间的链路关系根据真实的人际关系划分为:血缘关系、地缘关系、业缘关系、趣缘关系以及姻缘关系五类,并分析影响关系强度的因素。此外,引入流量关系网络建立了五种链路关系间的相互影响模型。在建模过程中,本文将代理行为分为认知、说服、决策、传播四个步骤,并采用基于规则的个体活动决策算法以及基于多态的传播决策算法对个体是否参加活动以及是否传播信息进行研究。最后通过对模型的有效性与正确性进行实证,研究不同因素对传播效果的影响。实验结果表明出度较大的节点作为发布者不仅能加强信息传播效果,而且对用户行为决策有一定的导向作用,信息质量、邻居节点态度以及信息发布频率等多种因素对传播以及预测效果均有不同程度影响。研究还发现:个体决策代价越大,个体决策过程越理性,则个体行为更容易预测。
[Abstract]:The research on the dissemination of information in social networks is not only of theoretical significance, but also of practical value. However, the complexity of link relationship, the diversity of information dissemination, and the heterogeneity of individuals all bring some difficulties to the development of this work. In recent years, the rapid development of computer technology provides a new way of thinking for the study of complex networks, and it is possible to reveal the law of information dissemination and predict the behavior of users. Based on the model of network structure, this paper uses agent-based modeling method to simulate the process of information transmission, summarizes the law of propagation, and predicts the behavior of users. Firstly, this paper abstracts the general rules of information dissemination and user decision from the application scene of information dissemination in social network. Secondly, the link vector model is used to model the social network. Online social network is the carrier of information dissemination. In this paper, the link relationship between nodes is divided into five categories according to the real interpersonal relationship: blood relationship, geographical relationship, business relationship, interest relationship and marriage relationship, and the factors influencing the relationship strength are analyzed. In addition, the interaction model of five kinds of link relationships is established by introducing the traffic relation network. In the process of modeling, the agent behavior is divided into four steps: cognition, persuasion, decision-making and communication. The rule-based individual activity decision algorithm and the polymorphic communication decision algorithm are used to study whether the individual participates in the activity and whether the information is transmitted. Finally, through the validity and correctness of the model, the influence of different factors on the communication effect is studied. The experimental results show that the node with relatively large degree can not only enhance the effect of information dissemination, but also have a certain guiding effect on the decision-making of user behavior, and the quality of information. Many factors, such as neighbor node attitude and information release frequency, have different effects on propagation and prediction effect. It is also found that the higher the cost of individual decision-making, the more rational the individual decision-making process is, and the easier it is to predict individual behavior.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5;G206

【参考文献】