A Study of Building Learning Expert System based on Ontology ─ a Case for Enterprise Network Diagnosis

• Main Authors: Po-Yen Pan, 潘柏延
• Other Authors: Ruey-Key Chiu
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/82073999028940827510

碩士 === 輔仁大學 === 資訊管理學系 === 93 === This paper presents a research attempting to add ontology elements in expert system, and hope to use ontology knowledge presentation – OWL (Ontology Web Language) as the resource of the domain knowledge of fact base in expert system. To stress the hierarchical relationship between knowledge, furthermore, make it possible to share and reuse the knowledge between different platforms (systems). This research also bases on the characteristics of decision tree. Taking advantage of ontology which is good at presenting hierarchical relationship conceptualization be diagnosing on nodes attribute of decision tree. Based on this idea, the design of the rule mechanism of decision tree can have the capability of self-learning. It can add the diagnostic basis and results produced by inference calculation and user’s knowledge (new fact) in the inference process to the fact base without altering rules and the inference system while not affect the normal operation of the system. So, the goal of expending knowledge and enhance inference accuracy and completeness and be effectively achieved. This research finds that the data (instance, attribute) in knowledge base can be used in rules. Because of rules on edition combine with ontology knowledge base, it makes instances and attributes which were defined by ontology can be reused on rules edition, thus achieves the objective of share and reuse. The decision trees of rules mechanism not only provides problem diagnosis, but promotes system inference efficiency by collection the knowledge feedback continually. We believe the inference mechanism which combines ontology on expert system presented in this research can be used as a referential model on different domain.