用户行为记忆性及其驱动机制研究

发布时间：2018-02-27 05:23

本文关键词： 人类行为动力学记忆模型记忆范围自激点过程学科背景　出处：《中北大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着信息技术的发展,越来越多的人类行为数据被记录,这有助于更加准确地统计与分析人类的行为规律,对舆情监控和引导、网络安全防护等都有重要的现实意义。为了进一步揭示用户在访问社交网络时的行为模式以及其规律性,本文对国内某高校师生用户访问社交网络的时间特性进行了分析,研究了用户行为的记忆性和活跃性,并揭示了影响用户行为的内在机制。主要的研究内容及成果包括以下几个方面:(1)采用自相关函数对用户的记忆性进行量化,从个体和群体层面分别研究了用户时间间隔的分布特征,并对用户相邻时间间隔的相关性进行了研究。研究结果表明,时间间隔服从幂律分布,且时间间隔的比率服从双帕累托分布,证实了用户行为中记忆性的存在。(2)分析了用户记忆性和活跃性之间的关系,并采用自激点过程理论刻画了用户记忆的变化过程。研究结果表明,用户的记忆效应取决于过去事件在当前时刻所产生的累计效应,且每一个事件产生的记忆性的长度都是有限的,该记忆长度的值不固定。(3)研究了不同学科背景的用户登陆行为之间的差异,并根据网站主要功能的不同将所研究的网站分为兴趣型网站和任务型网站,分别比较了兴趣和任务两种驱动因素对用户的不同影响。研究结果表明,学科背景在一定程度上会影响用户的活跃性;与任务驱动型网站相比,用户更倾向于访问兴趣驱动网站,同时用户的在线行为表现出了明显的节律性;由此建立了受学科背景和节律性影响的兴趣变化模型,且模型仿真结果与数据分析结果相一致。
[Abstract]:With the development of information technology, more and more human behavior data are recorded, which is conducive to more accurate statistics and analysis of human behavior, monitoring and guiding public opinion. Network security protection has important practical significance. In order to further reveal the behavior pattern and regularity of users visiting social networks, this paper analyzes the time characteristics of users visiting social networks in some colleges and universities in China. This paper studies the memory and activity of user behavior, and reveals the internal mechanism of influencing user behavior. The main research contents and results include the following aspects: 1) the memory of user is quantified by autocorrelation function. The distribution characteristics of user's time interval are studied from the individual and group levels, and the correlation of the user's adjacent time interval is studied. The results show that the time interval is distributed according to the power law. And the ratio of time interval from double Pareto distribution confirmed the existence of memory in user behavior. The self-excited point process theory is used to describe the changing process of user memory. The results show that the memory effect of the user depends on the cumulative effect of the past events at the present moment. And the length of memory generated by each event is limited, and the value of the memory length is not fixed. According to the main functions of the website, the website is divided into interest site and task site. The influence of interest and task on users is compared. Compared with task-driven websites, users tend to visit interest driven websites, and their online behavior shows obvious rhythm. The model of interest variation influenced by the background and rhythm of the subject is established, and the simulation results are consistent with the results of data analysis.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5;TP393.09

【相似文献】