Assistance in the management of rule sets for rule-based expert systems
(2023) The 35th International Conference on Software Engineering and Knowledge Engineering — Location: Hybrid, San Francisco, USA (1.July.2023)
Files
SEKE2023_REBS_rule_set_management.pdf
Open Access - Adobe PDF
- 290.66 KB
Details
- Authors
- Abstract
- Rule-based expert systems (RBES) use knowledge about a specific topic, represented as rules, in order to solve particular problems that would otherwise require a human expert. The creation and maintenance of the rule sets used come with several challenges in order to guarantee that they remain free of any error, which could reduce performances or lead to erroneous results. In this paper, we present a methodology to provide an automated assistance for domain experts creating and maintaining rules for RBES. This assistance takes the form of automated detection of relationships between rules that can lead to redundancies or conflicts. By reducing the weight borne by the human experts in the verification of the rules, it reduces the chance of errors, which helps increase the relevance of the rule set. To complete the theoretical methodology, we have implemented a functional prototype allowing for the management of a rule set and the visual highlight of redundancies and potential conflicts. Our approach, developed in the context of a case study, can be used for rules in any domain as long as they can be described in the same format. The approach can also be extended or modified to account for other types of relationships between rules.
- Affiliations
Citations
Legast, M., & Legay, A. (2023). Assistance in the management of rule sets for rule-based expert systems. International Conference on Software Engineering and Knowledge Engineering. Proceedings, 367-372. https://doi.org/10.18293/seke2023-189 (Original work published 2023)