Automatic reverse engineering extraction of ERC + from relational databases. (c1998)

Abstract

A reverse engineering method is presented and implemented in this work to extract a conceptual schema from a relational database. In this work we extend the extraction method presented in Andersson's paper entitled "Extracting an entity relationship schema from a relational database through reverse engineering". Such a method depends on the analysis of the database extension as well as the analysis of data manipulating statements, SQL statements, in the source code of a database application that uses a Relational Database Management System. Attributes representing references between tables in the relational schema are determined by an analysis of join conditions in all queries of the database. Candidate keys are determined with the help of the referencing attributes in their corresponding relations. The analysis of the database extension will determine whether the proposed keys are actually primary keys or not. Also, database extension helps in determining the inclusion dependencies between tables. When ambiguity occurs concerning candidate keys, users can interact with the system to clarity such ambiguity. This approach makes it possible to efficiently construct a conceptual schema, ERC+, from only elementary information.