Integrity Constraints Revisited
Robert Demolombe, ANDREW J. I. JONES · Logic Journal of IGPL · 1996
When computerised databases began to be used in business applications, there naturally arose a concern with how the integrity the stored data could be checked. This task was irnplemented by large sets of programs, depending on each par ticular application. Relational DBMS have significantly improved the situation in supporting autornated integrity checking, based on sorne particular kinds of Integrity Con straints (ICs). Despite this significant practica! irnprovement rnany researchers realized that the role and status of Integrity Constraints should be defined more precisely. This question is presently more important since Deductive DBMS pro totypes have been implemented, and the justification of the distinctions between the role played by the rules that are used to derive answers, and the role played by the rules that are used to check updates, also needs to be clarified. There are many papers in the literature that discuss such issues as: should ICs be characterized in terms of syntactical criteria like: atomic facts vs. general rules, ar Horn clauses vs. non Horn clauses?, How are IC violations to be formally characterized? What is the intuitive interpretation of IC violations? What is an ICs epistemic status?. The objective of this paper is to try to provide answers to such kinds of questions. We have tried to understand what the status and role of Integrity Constraints are, and we propose formal definitions and properties to represent our interpreta.tion. However, algorithms for integrity checking, and any consid erations of performance, are outside the scope of this paper.