项目名称: 在线交易行为实证与建模及虚假交易识别
项目编号: No.61304150
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 王澎
作者单位: 杭州师范大学
项目金额: 22万元
中文摘要: 本项目利用了复杂系统的研究方法,结合人类动力学的研究与复杂网络的建模思路对以淘宝为代表的在线交易系统中的用户行为进行实证与建模分析,并尝试据此建立一套基于用户行为特征的虚假交易识别算法。首先对交易系统的买卖双方的个体行为进行实证分析,重点利用人类动力学分析方法,研究其行为间隔分布,行为相关性, 突发性等动力学特征;从复杂系统与统计物理的角度实证研究买卖群体统计特征,包括相关统计量的分布特征,系统熵;特别是利用三部分图的方法对买卖双方交易动力学建模;基于以上建模与分析,找到能更好区分真实买家与虚假交易买家特征量或是阈值,通过买卖双方的关系网络以及买家间的团簇关系来定位抄信团伙,进而开发出更好的虚假交易识别算法,为净化中国电子商务环境做贡献。
中文关键词: 人类动力学;超图;虚假交易;复杂网络;
英文摘要: This project will study systematically and allsidedly user's behavior of on-line transactions systems like "taobao.com" basing on complex system theoy, the method of human dynamic and the modeling thoughts of complex network, and try to develop a new manipulative behavior detection system. First, the individual behaviors of buyers and sellers will be analyzed empricially, using the way from human dynamic, to abstract common rule on the user behavior in on-line transactions systems, for example: the distribution of inter-event times; from the angle of complex system and statistical physics, study the group structure charater, such as the distribution characteristics of relevant quantity, entropy;using the modeling framework of triple graph, study the dynamic of on-line trading; basing on the analysis above, try to find difference between the normal users and the cheaters;by means of triple graph find the group of the customers who provide manipulative ratings , bulid up a new manipulative behavior detection system. we hope this project can approach the application research of complex system theoy and increases prosperity of internet economy.
英文关键词: human dynamic;supergraph;manipulative behavior;complex network;