王怡斌

Yi-Bin Wang

碩士論文 (2006)

以本體論為基之分散式案例推理機制開發

Development of Mechanism for Ontology-Based Distributed Case-Based Reasoning

關鍵字 Keywords

本體論, 類神經網路, 案例式推理, 知識擷取

Neural Network, Knowledge Retrieval, Case-Based Reasoning, Ontology

摘要

在「知識經濟時代」與「分散式產業型態」中，企業知識資產的獲得已不再侷限於企業本身。為了支援分散式的產業環境，分散式案例推理勢必在跨企業的經驗與知識分享中扮演不可或缺的角色。目前在分散式案例推理的研究，多針對同一系統內之分散式案例作搜尋，且均依據事先設計之領域知識標準來實現知識分享，而系統間的知識分享卻未被重視。另外傳統的案例式推理系統，多僅提供使用者近似的案例，無法依使用者之需求進行調適。本研究主要目的在設計一分散式案例推理系統架構，並應用本體論工程技術來解決分散式案例異質與案例調適的問題，再利用本體論特性與多階段之演算法發展一「本體論為基之分散式案例推理機制」，使能依使用者之描述，從分散式歷史案例儲存庫中找尋到最近似之歷史案例，再經由領域本體論概念與概念之間的關聯與限制，導出案例調適的法則，再對案例進行調適，以協助使用者獲得適當之案例。

Abstract

Due to the advent of knowledge-based economy and distributed enterprises, the enterprises get the knowledge not only from themselves but also others. In order to support knowledge integration in the distributed enterprises, the distributed case-based reasoning systems（DCBRs） plays an important role in the knowledge and experience sharing. Up to the present, researches on DCBRs focus merely on retrieving cases in the same system and using a pre-defined standard of domain knowledge for knowledge sharing, but the needs of sharing knowledge among heterogeneous CBR systems have not been considered. In addition, traditional CBR systems only provide similar cases without performing case adaptation. The objective of the research is to develop a mechanism for ontology-based distributed case-based reasoning using characteristic of ontology and a proposed multistage algorithm. This thesis proposes a distributed CBR system architecture and uses ontology to solve the semantic mismatch problems between heterogeneous cases as well as the problems of case adaptation without involvement of domain experts. The results of this study will enable heterogeneous knowledge retrieval in distributed enterprises and thus facilitate knowledge sharing.