基于链接的作者重名处理方法研究与应用  

A Link-based Method for Name Disambiguation and its Application

作　　者：吴斌[1] 徐超群[1] 王文彬[1] 吴巍[1] 

机构地区：[1]北京邮电大学智能通信软件与多媒体北京市重点实验室,北京100876

出　　处：《计算机科学》2008年第3期197-199,共3页Computer Science

基　　金：国家自然科学基金项目(60402011);国家科技支撑计划项目(2006BAH03B05)

摘　　要：本文提出了一种适用于中文文献索引数据的实体解析方法。区别于新近的属性+链接结构的聚类方法,本方依据实际问题的特点采用了先属性匹配,然后基于文献合作网络的结构解析的策略。该方法在真实的大数据量文献索引数据上运行获得了良好的效果,并已经运用于数字图书馆的检索系统。The research presents an algorithm that is applicable to name disambiguation of Chinese literature digital library. Differ from the clustering method which considered the node attribute and link structure simultaneity, differ from the state-of-the-art LDA-ER method which employ the LDA model to resolute entities, differ from the DistQC model for resolving name disambiguation, our dedicated algorithm firstly makes an attribute similarity analysis and then detects the reference-entity relationship by considering the links of co-authorship network and the collective relations among the co-authorship network. The algorithm performs better on two datasets from a digital library and has a wide application in digital library.

关 键 词：实体解析 重名分析 信息检索 基于链接的分析 

分 类 号：TP393[自动化与计算机技术—计算机应用技术] F203[自动化与计算机技术—计算机科学与技术]