Representing and Reasoning About Mappings Between Domain Models

Abstract

Mappings between disparate models are fundamental to any application that requires interoperability between heterogeneous data and applications. Generating mappings is a labor-intensive and error prone task. To build a system that helps users generate mappings, we need an explicit representation of mappings. This representation needs to have well-defined semantics to enable reasoning and comparison between mappings. This paper first presents a powerful framework for defining languages for specifying mappings and their associated semantics. We examine the use of mappings and identify the key inference problems associated with mappings. These properties can be used to determine whether a mapping is adequate in a particular context. Finally, we consider an instance of our framework for a language representing mappings between relational data. We present sound and complete algorithms for the corresponding inference problems.