A Framework for Complete Automation of Database Rengineering Process

Abstract

Database reengineering consists of forward and reverse engineering. Database reverse engineering (DBRE) process is helpful in maintenance, extension and migration of legacy databases. Also, it can significantly help to reconstruct conceptual model, from modern databases which grow in ad-hoc manner. Current DBRE processes focus on retrieving ER diagram from databases, whereas our DBRE approach generates AER diagram. AER is an extension of an ER diagram accommodating Functional Dependencies as its integral part. This not only facilitates construction of conceptual model out of databases as a part of DBRE, but also allows complete automation of normalization of extracted conceptual schema to take care of anomalies. Thus, Our DBRE approach is an amalgamation of database schema and functional dependency extraction techniques. Database schema extraction algorithm developed in the paper, allows unconstrained data and it is completely automatic, meaning does not require any human interaction since mapping rules provide straightforward 1:1 mapping which results in unambiguous output. The FD extraction algorithm developed in the paper, called FD_COVER, retrieves optimum set of FDs and has the least time complexity compared to existing FD extraction algorithms. The approach results in a framework; we call it as AER IDE, which fully automates database reengineering process.