Resolving constraint conflicts in the integration of entity-relationship schemas

Abstract

In this work, we address the problem of constraint conflicts while integrating the conceptual schemas of multiple autonomous databases modeled using the Entity-Relationship (ER) approach. This paper presents a detailed framework to resolve three types of constraint conflicts, domain constraint conflicts, attribute constraint conflicts and relationship constraint conflicts. There are two types of domain constraint conflict, convertible and inconvertible. We distinguish two types of convertible domain constraints conflict, reversible and irreversible, and present an algorithm to resolve domain constraint conflicts. We identify six factors that can contribute to conflict in attribute constraints: imprecise constraint design, domain mismatch, incomplete information, imprecise semantics, value inconsistency and set relation between object types. In relationship constraint conflict resolution, we examine the set relation between equivalent relationship sets and the functional dependencies that hold in these relationship sets. Our conflict resolution approach does not assume that equivalent entity types or relationship sets in two schemas model exactly the same set of instances in the real world. Furthermore, our approach enforces the most precise constraints and enables the retrieval of all the data in the local databases via the integrated schema.