Knowledge Base for Database Design

1. INTRODUCTION This paper describes the database design knowledge of a computer-assisted database design tool, called the Intelligent Database Design Kit (IDDK): useil to develop database sy&ms. The IDDK knowledge base 1s stored in the form of a database dictionary based on an extended entity-relationship-ataibute model. The database dictionary is central to the knowledge engineering approach of the IDDK implementation strategy. The knowledge-based dictionary is a repository of meta-data (atomic facts, concepts. definitions, inference rules, heuristics) for a database system development. IDDK tools are developed using, to a large extent, semantic models and artificial intelligence techniques. The paper introduces computer-assisted extensions to Chen's entity-relationship model, in particular, recursive, subset, generic, and nested relationship concepts. An architecture and framework for intelligent database design is presented. The database design knowledge is divided into a declarative part (containing a description of entities, relationships and attributes) and a procedural part (defining design operations). While driving the IDDK tools, the knowledge base maintains basic consistency checks. In fact, the same methodology that supports the database design activities is used to implement the knowledge base. The IDDK interface to the knowledge base editor is based on a standard Macintosh interface extended with database design-oriented pallette icons and graphical editing tools. Retrospectively, database design has relied ujton human experience and judgement rather than mechanistic algcnithmS. The design task is usually performed by vltperts who obtain information about the users' needs through interviewing, examining existing documentation, and other traditional means. To this extent, the current approach suffers from two weaknesses: (1) it requires the use of a scarce resource-the expert database designer, and (2) the designer's knowledge of the application is necessarily second-band and some of its intricacies ,are likely to be overlooked. In our research, we respond to these problems by providing: (1) an expert system based set of tools, (2) a large lmowledge base ;IS a repository for information about the application and design activities. The Intelligent Database. Design Kit (IDDK) is a knowledge engineering project which integrates database technology. software engineering, and knowledge-based techniques. A key concept in this research is the use of a semantically extended conceptual model (entity-relationship (Chcn, 1976)) as the underlying formalism of the database design knowledge base, which in turn drives the IDDK tools. The paper extends the fonhcoming ANSI and IS0 standards for Information Resource Dictionary System IRDS (Dolk and Kirsch II, 1987; Goldfine, 1985). The extension is threefold: …