Schema mapping generation in the wild
作者：
Highlights：
• A dynamic programming algorithm searches a space of mappings for a set of relations.
• The source relations are merged on the basis of relational metadata and profile data.
• An inference mechanism for propagating profile data without data materialization.
• A fitness function for comparing candidate mappings using inferred profiling data.
• A method that populates a multi-relation target schema with constraints.
• An empirical evaluation on real-world and (benchmark) synthetic datasets.
摘要
•A dynamic programming algorithm searches a space of mappings for a set of relations.•The source relations are merged on the basis of relational metadata and profile data.•An inference mechanism for propagating profile data without data materialization.•A fitness function for comparing candidate mappings using inferred profiling data.•A method that populates a multi-relation target schema with constraints.•An empirical evaluation on real-world and (benchmark) synthetic datasets.
论文关键词：0000,1111,Mapping generation,Profiling data,Dynamic programming
论文评审过程：Received 23 April 2021, Revised 17 August 2021, Accepted 27 September 2021, Available online 14 October 2021, Version of Record 22 October 2021.
论文官网地址：https://doi.org/10.1016/j.is.2021.101904
