Experimental Performance Evaluation of Uncertain Ontology Representation Using Bayesian Model

Uncertain knowledge allows us to learn new knowledge and leads us to develop a new domain model. The medical domain is a very crucial area of research where a wrong decision taken by a physician can create a significant health issue. Based on patient sensations and certain conventional measurements, physicians attempt to identify possible diseases and doubts. But they fail to do so and deal with uncertainty. Therefore, we need to be certain of the method of any new observation. The Bayesian network (BN) model deals with uncertainty for better knowledge modeling and representation, respectively. This paper focuses on knowledge representation and the prediction of uncertain knowledge in domain ontology (Liver disorder) using the BN model. In this paper, we have evaluated the accuracy of each node and experimentally validated it. We have also compared our proposed model (BN Model) with Random Forest (RF), Hidden markov model (HMM), Support vector machine (SVM), K-NN, and Neural Network (NN) algorithms to check model efficiency to find its sensitivity, specificity, and accuracy. From the result, we found that our approach (BN Model) on the same dataset (Liver Disorder Ontology) outperforms in comparison to other modeling techniques