AbstractThe world in which we live is constantly changing and information systems must also evolve to cope with these changes. However, legacy systems, which are very large and complex, cannot easily evolve. In particular, the semantics of the logical database schema of these systems is often not documented. Therefore, the conceptual schema will have been lost over time. Unimplemented foreign keys are then lost, making the task of upgrading this database perilous. To compensate for this, this conceptual schema must be restored. Database reverse engineering provides tools for retrieving this lost information. Nowadays, locating implicit foreign keys in order to implement them is no longer a problem. However, the restoration of these relational constraints is not yet a subject of much study.
This thesis aims at proposing solutions to facilitate the implementation of these lost implicit foreign keys thanks to a method of evaluation of the risk of implementing a constraint through a calculation of the schema transformations necessary to its addition as well as the probability and the importance of the errors that it could generate at the application level.
|Date of Award
|7 Sept 2018
|Anthony Cleve (Supervisor)