FUTURE TRENDS IN EXPERT DATA BASE SYSTEMS - read and discuss expert DBMS evolution

Sep 17, 2026

Uploaded by Skyler

The paper examines how database management systems (DBMSs) are expected to evolve over the next several years to support expert database applications. It highlights research thrusts receiving insufficient attention, proposes architectural approaches such as loose and tight coupling between rule management and data management, and evaluates techniques for triggering rules from large rule sets. It concludes that building real-world expert systems requires far more than adding rule management to a DBMS, including challenging object management issues.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Architectures for Expert Data Base Applications

Examples of Expert Data Base Applications

Rule Firing and Technical Approaches

Real-World Construction and Object Management

FUTURE TRENDS IN EXPERT DATA BASE SYSTEMS Michael Stonebraker and Marti Hearst EECS Dept. University of California, Berkeley Abstract In this paper we discuss how we see the capabilities of DBMSs evolving over the next several years to meet the needs of expert data base applications. We also present some of the research thrusts which we see as important that appear to be receiving insufﬁcient attention in the research community. 1. INTRODUCTION We begin in Section 2 with an example application which we think characterizes the need for expert data base management systems. This application will set the context for the discussion in the rest of the paper. This paper then turns in Section 3 to the architectural alternatives that can be adopted for the construction of expert data base systems such as the one described in Section 2. We begin with loose coupling between a rule manager and a data manager. The serious ﬂaws in this architecture are then discussed followed by the possibility of tight coupling between rule and data management. We then point out that tight coupling will allow simple expert systems to be entirely implemented within the data manager while more complex ones can be written in a cooperative style whereby a data manager with an embedded inference engine and an expert system shell work cooperatively to implement an expert data base application. Then we turn in Section 4 to the two thrusts which research on DBMSs with an embedded rule system have taken. A collection of researchers have concentrated on the efﬁcient solution of rules systems that are recursive in nature. These naturally lead to the construction of query optimizers for a DBMS that can deal with recursive queries. On the other hand, other researchers have focused on more mundane rule systems where recursion is not present and have worked on efﬁciently discovering when one must "ﬁre" rules from a large rule set. We conclude Section 4 by discussing the importance of these two thrusts and their likely commercial exploitation. In Section 5 we discuss the possible technical approaches to deciding when to "ﬁre" rules from a large rule set. We discuss approaches based on theorem provers, indexing techniques and physical markers in the data base. We conclude that none are satisfactory alternatives, and we are enthusiastic about research on other possible mechanisms. Section 6 then concludes by indicating that the construction of real world expert systems entails much more than adding rule management to a DBMS. Speciﬁcally, expert data base applications seem to entail a collection of difﬁcult object management problems. In Section 6 we discuss a simple expert system which we have been working on and indicate why a conventional relational system with an inference engine is not adequate for our application. This research was sponsored by the U.S. Air Force Ofﬁce of Scientiﬁc Research Grant 83-0254 and the Naval Electronics Systems Command Contract N39-82-C-0235 2. AN EXAMPLE EXPERT DATA BASE APPLICATION Consider a computer program which is intended to give a person directions from any starting location in a particular geographic area to his destination. Such a program would be useful to car rental companies who must give directions from an airport location to the traveler’s ﬁnal destination. It would also be useful to route delivery vehicles in unfamiliar territory, etc. In fact, simple systems with the capabilities which we will describe are under development commerically at the present time. The intended program will run off a data base which consists of a street map of the area in question. For each adjacent pair of nodes, the data base must store: the travel time along the arc (perhaps by time of day) the legal house numbers along the arc whether it is a one-way street the presence of stop signs the road surface For a geographic area such as the San Francisco Bay area, there are (say) 5 million people, two million le

FAQs

What is the main focus of the paper on expert database systems?

The paper discusses how DBMS capabilities should evolve to meet the needs of expert database applications and identifies key research areas that may be underemphasized.

How does the paper compare loose coupling and tight coupling architectures?

It first considers loose coupling between a rule manager and a data manager, then discusses flaws in that approach, and argues that tight coupling can enable expert systems implemented within the data manager or in a cooperative style using an embedded inference engine.

Why are existing methods for deciding when to fire rules considered unsatisfactory?

The paper reviews approaches based on theorem provers, indexing techniques, and physical markers, then concludes none are fully satisfactory alternatives and calls for other mechanisms.

Sep 17, 2026

Uploaded by Skyler