Date of Award

1988

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Peter P. S. Chen

Abstract

There is a growing need to incorporate database integrity subsystems into large information systems in engineering design environments and real-time control and monitoring environments. The objectives of the integrity subsystem are to provide a user interface for constraint specification, to compile the specification into enforcement strategies, and to check data integrity at both compile-time and run-time. The approach proposed by this research is to develop the conceptual view of the database using the Entity Relationship Model (ERM). Users' queries and semantic constraints can be specified by an ER-based data language, the Applicative Data Language (ADL). Any ADL constraint specification is compiled into both a compile-time and a run-time checking strategy for enforcement. The integrity subsystem, then, automatically maintains the consistency of data whenever there is a change in the database state. The basic constructs of ADL are data structures, functions, and predicates. It takes advantage of the semantic clarification of objects and relationships in the Entity Relationship Model by doing, first, an object level computation and, then, a data element level computation. The object level computation determines how objects are associated with each other. The data element computation, on the other hand, examines the data values of those associated objects and derives new relations from these values. A semantic constraint, therefore, is formulated as a computation procedure that maps the current database state to a TRUE or FALSE value. The computational syntax of ADL allows us to compile directly each constraint specification into a transition digraph for compile-time constraint checking. This research proposes the incremental computation strategy for efficient run-time constraint checking. The objective of the strategy is to do run-time constraint checking without full evaluation of the database. The entire computation procedure centers around the user's update. It propagates the incremental changes along the transition digraph to infer the effect of the update upon the new truth value of the semantic constraint. This research concludes that ADL with its generality in semantic constraint modeling and its enforcement strategies at both compile-time and run-time is adequate as the architecture for an integrity subsystem supporting an Entity Relationship database.

Pages

219

Share

COinS