A Service-based Approach to Schema Federation of Distributed Databases

In the last few years, we have witnessed a rapid growth in distributed database processing. We consider the question of data integration: how we can integrate distributed schemas into a new one and query just that new schema without losing the ability to retrieve data from the original schemas. The area in which we try to answer that question is federated databases, where the original heterogeneous sources are assumed to be autonomously managed. Our approach is based on schema matching: the identification of semantic correspondences between elements of different schemas. We propose SASF, the Service-based Approach to Schema Federation, which is composed of three services: schema translation, schema matching, and schema mapping that are accessed through a user portal. Our approach exploits both schema-level and instance-level information to discover semantic correspondences, and is able to operate over a full range of matching cardinalities. A demonstration of the usability of SASF in a real-world scenario of federating astronomical databases is presented as a case study. The results, we believe, illustrate the potential of performing data integration through database federation. Edinburgh E-Science Technical Report EES-2006-01 Available from http://www.nesc.ac.uk/technical_papers/EES-2006-01.pdf Copyright © 2005 The University of Edinburgh, All rights reserved