维基百科词条编辑特性研究  

Research on Article Edit Characteristic in Wikipedia

作　　者：赵东杰[1,2] 郝黎[3] 李德毅[4] 王华[5] 何宇[1] 

机构地区：[1]装备指挥技术学院 [2]63628部队 [3]北京航空航天大学 [4]中国电子系统工程研究所 [5]中国航天员科研训练中心

出　　处：《计算机科学》2011年第B10期153-156,共4页Computer Science

基　　金：国家自然科学基金项目(69120912,61035004);国家973计划项目(2007CB310804)资助

摘　　要：针对维基百科词条编辑特性问题,以网络化数据挖掘思想方法为指导,对高质量维基百科词条进行文本分析,判断词条相邻版本间句子差异,以编辑者为节点,编辑者间编辑交互关系为连边,构建词条编辑交互网络,通过分析网络结构特征实现词条编辑特性分析。分析表明网络具有小世界特性,度分布与强度分布相似,具有较强正相关性,其累积分布与边权重分布服从幂律分布,节点度与聚集系数具有较强负相关性,最短路径长度分布与高斯分布相似,网络具有异配性和较弱的互惠性,编辑群体具有较强异质性、抱团性;深化了对词条编辑交互过程和群体智能的认识。Aiming at the problems of article edit characteristic in wikipedia,under the direction of the idea of networked data mining,featured articles in wikipedia were analysed by text processing to fiind the difference of sentence between adjacent versions,the article edit interaction networks were constructed,where the node is editor and the link is the edit interaction connection between editors,then the article edit characteristics in wikipedia were analysed by the empirical analysis of the nine networks.Results show that all networks have small-world properties,strong positive degree-strength correlation and negative degree-clustering coefficient distribution,their degree distributions are similar to strength distributions,their cumulate distributions and link weight distribution are power-law distribution,shortest path length distributions are similar to gauss distribution,all networks have degree and strength disassortativity and weak reciprocity,furthermore the edit collective have strong heterogeneity and community structure,which deepens the know-ledge of the process of article edit interaction and collective intelligence.

关 键 词：维基百科 词条编辑交互网络 网络化数据挖掘 群体智能 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]