Answers are grounded in this paper and cite their sources.
Knowledge engineering in Knowledge-Based Systems (KBS) still largely relies on manual approaches, which limit scalability, adaptability, and development efficiency. Meanwhile, data-driven approaches, such as Case-Based Reasoning (CBR), are typically used only as problem-solving mechanisms and have not been fully leveraged as primary sources for structured knowledge construction. This study aims to bridge the gap by developing a Knowledge-Based System Generator that automatically builds a knowledge base from case-based datasets. The proposed system employs the C4.5 algorithm to generate decision trees and diagnostic rules from historical data automatically. The generated knowledge is evaluated using several inference methods, namely the Matching Rule, Forward Chaining, Backward Chaining, Random Forest, and Jaccard Coefficient. The evaluation was conducted on ten disease datasets under three testing scenarios: testing on a dataset different from the training data, testing on the entire dataset, and testing on the same dataset for both training and testing. The experimental results show that the system achieves an overall average accuracy of 0.8604, with the best performance obtained using the Jaccard Coefficient, which reaches an average accuracy of 0.9145. In addition, the efficiency analysis indicates that the system can construct a compact, representative knowledge base with an average efficiency of 66.69%. These findings demonstrate that automatic case-based knowledge generation is feasible, effective, and generalizable. Therefore, this study contributes to shifting the KBS paradigm from manual knowledge engineering to adaptive and scalable data-driven knowledge generation.
No takes yet. Share an insight, caveat, or question.
Delima et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: