Converting Legacy Relational Database into XML Database through Reverse Engineering

XML (eXtensible Markup Language) has emerged and is being gradually accepted as the standard for data interchange over the Internet. Since most data is currently stored in relational database systems, the problem of converting relational data into XML assumes special significance. Many researchers have already done some accomplishments in this direction. They mainly focus on finding XML schema (e.g., DTD, XML-Schema, and RELAX) that best describes a given relational database with a corresponding well-defined database catalog that contains all information about tables, keys and constraints. However, not all existing databases can provide the required catalog information. Therefore, these applications do not work well for legacy relational database systems that were developed following the logical relational database design methodology, without being based on any commercial DBMS, and hence do not provide well-defined metadata files describing the database structure and constraints. In this paper, we address this issue by first applying the reverse engineering approach described in (Alhajj 2003) to extract the ER (Extended Entity Relationship) model from a legacy relational database, then convert the ER to XML Schema. The proposed approach is capable of reflecting the relational schema flexibility into XML schema by considering the mapping of binary and nary relationships. We have implemented a first prototype and the initial experimental results are very encouraging, demonstrating the applicability and effectiveness of the proposed approach.