Multi-Path Reasoning in a Database

  • Authors:
  • Dariush Navabi;Lois M. L. Delcambre

  • Affiliations:
  • Indotronix International Corporation, P.O. Box 6973, Chandler, Arizona 85246. E-mail: navabi_d@srm1.sat.mot.com;Oregon Graduate Institute of Science and Technology, Computer Science and Engineering Department, P.O. Box 91000, Portland, Oregon 97291. E-mail: lmd@cse.ogi.edu

  • Venue:
  • Journal of Intelligent Information Systems
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

Rules have been used in a database context for several purposes:deductive database queries, active database triggers, and productionsystem programs. Exploring the search space for non-deterministicrule programs, however, has generally been available only in largemonolithic systems intended for artificial intelligence applications.The goal of this research is to provide multi-path reasoning fornon-deterministic rule programs in a database environment. Themotivation for this work includes the increasing use ofdatabase-style triggers to assist data-intensive applications, e.g.,the human genome project and the Intelligent Integration ofInformation (I^3) Program.A non-deterministic rule program is one where there is more than oneterminal state. Such programs are generally not consideredappropriate for database queries where the focus is on rules programsor techniques that guarantee a unique fixed point. Butnon-deterministic programs are commonly used for various heuristicsearch problems such as the traveling salesperson problem. With anon-deterministic program, it is particularly important to beflexible about the order in which the search space is exploredbecause exhaustive search is generally not feasible. Thecontributions of this research are: the representation of themulti-path search tree within a database and the ability for theproblem solver to control the search process for a non-deterministicrule program.