Semi-automatic schema integration in Clio

Abstract

Schema integration is the problem of finding a unified representation, called the integrated schema, from a set of source schemas that are related to each other. The relationships between the source schemas can be represented via correspondences between schema elements or via some other forms of schema mappings such as constraints or views. The integrated schema can be viewed as a means for dealing with the heterogeneity in the source schemas, by providing a standard representation of the data. Schema integration has received much of attention in the research literature and still remains a challenge in practice. Existing approaches require substantial amount of human feedback during the integration process and moreover, the outcome of these approaches is a single integrated schema. In general, however, there can be multiple possible schemas that integrate data in different ways and each may be valuable in a given scenario.