Course Indicators for Forward Logical Inference upon List Variables
Abstract
Introduction: Flexible discrete systems (expert systems, information systems, etc.) often face the problem of a processing halt, i.e. detection of a processing step demanding a change in the system operation mode (for example, generating a failure signal or inverting the direction of logical inference). This paper introduces some universal heuristic indicators to control the course of logical inference in such systems. These indicators can be specified for real data types and used for the estimation of success in advancing to the inference goal in order to make a decision regarding the termination or prolongation of the inference. Results: A technique is proposed to control forward logical inference in flexible discrete intelligent systems, using heuristic set-theoretic indicators to analyze the current subsets of possible, false and true values of variables, terminating an inference chain if it looks unpromising or containing a conflict. A chain is unpromising when there are no ways to narrow down a current subset of possible values for the goal variable within this chain. An example of a conflict is obtaining an empty set of possible values for a variable. The developed method uses integrated output management ratings success inference procedures according to the criteria of "common sense", which include coefficients of confidence, calculation of trust and confidence (chance) in respect to hypotheses considering used and not yet used evidences, JSM method ideas, etc. Practical relevance: In flexible discrete systems, the introduced rules of productions application and empirical indicators of logical inference course let you estimate the progress in moving to the inference goal and make a decision whether to terminate or to resume the inference chain.Published
2015-02-20
How to Cite
Fridman, A., Kulik, B., & Kurbanov, V. (2015). Course Indicators for Forward Logical Inference upon List Variables. Information and Control Systems, (1), 29-34. https://doi.org/10.15217/issn1684-8853.2015.1.29
Issue
Section
Information and control systems