ASSESSING THE MERITS AND LIMITATIONS OF A MULTILAYER TREE STRUCTURE BELIEF RULE-BASED EXPERT SYSTEM

ABSTRACT:

The Belief Rule Base (BRB) expert systems have emerged as a prominent approach in expert knowledge representation and reasoning. BRB systems offer interpretability through easily understandable IF-THEN rules, facilitating validation by domain experts and enabling insight into the inference process. However, conventional BRB systems encounter challenges such as: combinatorial explosion, suboptimal parameter learning, and a lack of self-tuning mechanisms. The Multilayer Tree Structure BRB (MTS-BRB) model presents a new approach to tackle the combinatorial explosion problem associated with the conventional BRB model and improve the model inference mechanism. This paper provides a comprehensive assessment of the merits and limitations of the MTS-BRB expert systems. This study has shown that the MTS-BRB model has made notable improvements on the conventional BRB model however, it has not completely addressed all the former issues, and it also presents its challenges.

Keywords: Explainable Artificial Intelligence, Expert Systems, Belief Rule Base, Machine Learning