Semantic adaptation of schema mappings when schemas evolve

Abstract

Schemas evolve over time to accommodate the changes in the information they represent. Such evolution causes invalidation of various artifacts depending on the schemas, such as schema mappings. In a heterogenous environment, where cooperation among data sources depends essentially upon them, schema mappings must be adapted to reflect schema evolution. In this study, we explore the mapping composition approach for addressing this mapping adaptation problem. We study the semantics of mapping composition in the context of mapping adaptation and compare our approach with the incremental approach of Velegrakis et al [21]. We show that our method is superior in terms of capturing the semantics of both the original mappings and the evolution. We design and implement a mapping adaptation system based on mapping composition as well as additional mapping pruning techniques that significantly speed up the adaptation. We conduct comprehensive experimental analysis and show that the composition approach is practical in various evolution scenarios. The mapping language that we consider is a nested relational extension of the second-order dependencies of Fagin et al [7]. Our work can also be seen as an implementation of the mapping composition operator of the model management framework.