基于代理的社交网络用户行为建模与实验分析

发布时间：2018-07-02 20:55

本文选题：用户行为建模 + 决策机制　；参考：《南京邮电大学》2017年硕士论文

【摘要】：社交网络用户行为建模旨在最大限度模拟人的真实行为效果并分析其中的行为规律,进而对某些行为现象给出合理的解释。然而,早期的用户模型仅将用户当作社交网络中的一个节点,基于图论所构建的用户影响力模型或是基于统计学方法构建的行为动力模型都没有深入探究行为产生的内部决策机制,另外,有些模型虽然涉及到内部决策机制,但是不够全面,难以做到拟人决策。本文从微观个体出发,构建了延展闭环心理过程决策模型,并将模型应用于抽象的社会阶层网络中,采用基于代理的建模方法,构建了用户状态迁移模型。本文首先基于应用场景提出决策模型的基本框架,并在此基础上细化了决策的心理过程,进而构建情感激励模型,双通道决策模型,意图波动模型和体验反馈模型四个过程模型。随后,本文在社会阶层网络中构建用户的状态迁移模型,从决策的角度研究社会阶层流动现象。最后,本文结合自驾游社群活动数据,验证了心理过程模型的优越性,并利用Netlogo软件进行实验分析,验证了状态迁移模型的有效性和正确性。本文得到以下结论:在决策模型中构建完善丰富的心理过程能够提高模型的预测准确性,并且在间隔时间较长的决策情境下,用户往往会改变自己的原意图。另外,本文提出的状态迁移模型得到了与实际情况较为拟合的阶层状态分布曲线。研究还发现:用户迁移的内部动机越大,则用户越容易发生状态迁移。
[Abstract]:Social network user behavior modeling aims at maximizing the effect of human behavior and analyzing the behavior rules, and then giving a reasonable explanation for some behavior phenomena. However, the early user model only regards users as a node in the social network, and the user influence model based on the graph theory or based on statistics is based on statistics. In addition, some models are not fully comprehensive and difficult to make human decision making. In this paper, a decision model of the extended closed loop psychological process is constructed and the model is applied to an abstract society. In the hierarchical network, the user state migration model is constructed by using the agent based modeling method. Firstly, based on the application scenario, the basic framework of the decision model is proposed. On this basis, the psychological process of decision making is refined, and then the emotional incentive model, the dual channel decision model, the intention fluctuation model and the experience feedback model are four over. Then, this paper constructs a user's state migration model in the social stratum network and studies the social stratum flow phenomenon from the decision point of view. Finally, the superiority of the mental process model is verified by the data of the self driving community activity, and the effectiveness of the state migration model is verified by the Netlogo software. The following conclusions are obtained: the construction of a perfect and rich psychological process in the decision-making model can improve the prediction accuracy of the model, and the user often changes its original intention map in a decision-making situation with a long interval. In addition, the state migration model proposed in this paper has got a more fitting class with the actual situation. The state distribution curve also shows that the greater the internal motivation of user migration, the more likely users are to migrate.
【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

【参考文献】