数据库模式匹配:一种查询逆向工程方法  

A QUERY REVERSE ENGINEERING APPROACH TO DATABASE SCHEMA MAPPING

作　　者：刘履宏 何震瀛[2] 荆一楠[2] Liu Lühong;He Zhenying;Jing Yi'nan(Software School,Fudan University,Shanghai 201203,China;School of Computer Science,Fudan University,Shanghai 201203,China)

机构地区：[1]复旦大学软件学院,上海201203 [2]复旦大学计算机科学技术学院,上海201203

出　　处：《计算机应用与软件》2020年第10期20-26,144,共8页Computer Applications and Software

基　　金：国家重点研究计划项目(2018YFB1004404)。

摘　　要：传统的基于相似度的数据库模式匹配方法不仅费时费力,而且要求用户对数据源非常了解。为此提出一种基于查询逆向工程的模式匹配方法。根据用户提供的符合目标模式的实例数据以及源数据库,计算出源数据库生成给定实例的查询,在实例数据列和源数据库列之间建立映射关系,帮助用户自动完成数据库模式匹配。针对处理多行实例数据提出基于重用和基于列的优化方法,避免了冗余查询的生成和验证,提升了匹配效率。在IMDB数据集上的实验结果表明,与基础方法相比,基于重用的模式匹配方法和基于列的模式匹配方法的运行效率分别提升了3~5倍和10倍。Traditional similarity-based database schema mapping methods are time-consuming and laborious,and they require users to have detailed knowledge of data sources.In this respect,this paper proposes an automated schema mapping method based on the query reverse engineering techniques.According to user-provided instances and the source database in accordance with the target schema,this method identified the mapping relationship between the columns of instances and the source database columns by rediscovering the queries that generate the given instances from the source database to help users automatically perform the schema mapping process.It avoided the generation and verification of redundant queries when dealing with multiple instances for reuse-based schema mapping method and column-based schema mapping method to improve the schema mapping efficiency.The experimental results on the IMDB dataset show that the efficiency of the reuse-based method and the column-based method are improved by 3~5 times and 10 times respectively than that of the baseline method.

关 键 词：数据库 模式匹配 查询逆向工程 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]