Abstract

An important issue in modern information systems and e-commerce applications is providing support for inter-operability of independent data sources. A broad variety of data is available on the web in distinct heterogeneous sources, stored under different formats: database formats (e.g., relational model), semi-structured formats (e.g., DTD, SGML or XSD schema), scientific formats, etc. Integration of such types of data is an increasingly important problem. Nonetheless, the effort involved in such integration, in practice, is considerable: conversion of database schemas from one format to another requires writing and managing complex data transformation programs or queries. Realizing the importance of converting structured schemas to semi-structured schemas and vice versa, we have developed a method that successfully performed the conversion. Our approach benefits highly from the reverse engineering of structured or semi-structured schema into a ContextMap representation and leads to both identifying and understanding all components of an existing database system and the relationships among them. We have successfully developed a prototype, called CODAX, for the schema conversion process. While more sophisticated techniques are required in this context, we believe the ideas proposed in this work lend themselves to useful analysis and tracing tools for schema conversion