一种模式匹配和实体统一相互促进的方法  

An Improvement Method of Schema Matching and Entity Resolution

作　　者：潘峰[1] 李庆忠[1] 董永权[1] 

机构地区：[1]山东大学计算机科学与技术学院,济南250101

出　　处：《计算机与数字工程》2009年第9期4-6,19,共4页Computer & Digital Engineering

基　　金：国家自然科学基金(编号:60673130);山东省自然科学基金(编号:Y2007G38);山东省科技攻关计划(编号:2008GG30001005);山东省科学技术发展计划(编号:2007GG1QX01036);山东省博士后创新项目专项资金(200703084)资助

摘　　要：在数据库研究领域,模式匹配和实体统一是被广泛关注的两个方向。随着对Web数据集成需求的增长,无论是在模式和实体层次,研究这两方面问题是很有实际意义的。当前的研究大多针对两项任务的其中之一。在文章中,基于模式匹配促进实体统一的新思路,提出了一种同时解决这两项任务的方法,实现了它们之间的相互促进机制。在现实的Web异构数据源场景中应用该方法,得到的查准率和查全率都很高,证明了该方法的正确性和有效性。Schema matching and entity resolution have been two topics widely studied in the field of database research. With the rising demand in the Web data integration, both in schema and instance level, the study of the two tasks is becoming more practical importance. Most current study efforts at resolving one of the two matching tasks. In this paper, based on the new ideas of schema matching benefit from entity resolution, we propose a method that simultaneously attacks these two tasks and achieves a kind of improvement between them. By applying our method to a realistic Web heterogeneous data source scenario, we show that precision and recall are both quite high, and show this method＇s correctness and validity.

关 键 词：数据集成 模式匹配 实体统一 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]