Exploration of rule-based knowledge bases: A knowledge engineer's support

摘要

Data exploration helps us understand the investigated reality in a faster and better way. In this paper, the data to be explored are domain knowledge bases with rules representation. The specificity of rule representation requires optimum selected analysis methods to provide useful new knowledge to both the knowledge engineer and the user of a decision support system with a rule-based knowledge base. Effective exploration of rule-based knowledge bases can be carried out through the creation of if... then clusters of rules and their representatives using hierarchical methods. This is a new and unique approach to the managing of domain knowledge bases as it facilitates the creation of cohesive and well-described clusters and the detection of rare rules (those dissimilar to other rules) while concurrently providing visualization of a knowledge base. In the experiments, four knowledge bases with a varying number of attributes and rules have been used. The knowledge bases have been explored using four different methods of determining clusters' representatives, four clustering methods and nine similarity measures. It turns out that each of the factors substantially influences to the size of the resulting clusters, the number of outliers and the occurrence frequency of overgeneral and overspecific rule clusters' representatives. (C) 2019 Elsevier Inc. All rights reserved.