A logic programming perspective on decision table theory and practice

摘要

A theoretical foundation is presented aimed at reconciling the theory and practice of decision table (DT) processing with a family of relatable methodologies represented by logic programming (LP), Prolog, and relational databases. Formal DT definitions, based on ones promulgated by Codasyl's Decision Table Task Group in 1982 and modified in the paper, promote a balanced emphasis on functional and relational representations established in the theory and in prototype implementations. Implementations with term data organizations (constant, variable and function) adhere to the functional definition of DT's and their origins in a larger DT processing context and potential automatic interconnections are delineated. Implementations with predicate data organizations adhere to the relational definition of DT's and illustrate the impact of relational level processing on DT processing; interconnections among these implementations are also stipulated. Use and performance characteristics for the various methods are mentioned and, in order to highlight the contributions to DT theory and methodology, the attractiveness of the perspectives gained from LP is summarized and demonstrated. First, there are important DT properties obtained at little cost in extra coding: capturing of precepts on ordering within DT's and don't-care, facilitating DT updating and integrity constraints, and expediting implementation of table systems. Then, novel means of simulating policy maps, wide ranging DT accessing mechanisms, and a connection to relational database theory display efficacy of the approach by demonstrating how certain very desirable processing goals can be readily achieved within the purview provided. Culminating remarks indicate how LP and DT's constitute mutually supporting methodologies, i.e., together they offer prospects for enhancement and automation of whole lifecycle schemes, including aids for developing programs written in Prolog.