陳怡燕

I-Yen Chen

碩士論文 (2006)

以本體論為基之知識整合機制研發

Development of an Ontology-based Knowledge Integration Mechanism

關鍵字 Keywords

知識整合, 本體論整合, 本體論, 本體論映對, OWL

OWL, ontology mapping, knowledge integration, ontology integration, ontology

摘要

知識經濟時代來臨，知識成為影響組織競爭優勢的關鍵要素。基於知識變化與更新的迅速，知識工作者所需的知識通常必須由不同之來源與管道取得，因此，整合相關領域知識提供使用者運用、分享與再利用，才能發揮知識之價值。然而「知識」的存在為一抽象的概念，不易明確的描述與定義，造成知識整合所面臨之瓶頸。 　　本體論可具體表達人類知識概念並呈現知識結構，為近來知識表達、知識工程等相關領域的新興理論與技術。然而知識之分析與界定因人而異，即使透過本體論模式加以表達，同一領域內之本體論仍會因使用者觀點的不同，產生概念表達時語意或結構的差異，造成本體論之異質性，因此知識整合之核心就是本體論整合。 　　本研究之目的在發展一「以本體論為基之知識整合機制」，包含「本體論映對模式」建構與「本體論相似度比對演算法」設計，透過本研究之機制可自動地找出本體論之間語意相關的部份，映對不同本體論之間相同的概念，實現異質本體論的整合，並藉由本體論整合達到知識的整合。 　　本研究之本體論映對方法除了運用本體論的類別與屬性等元素，還運用本體論之「內部關係」，以提升映對的準確度。整合後的本體論除了可實現知識整合外，還能運用在系統整合方面以提升系統間的互通性，並可利用本體論映對方法建構出更健全的語意網以提升知識搜尋、萃取及推論的準確性。

Abstract

Knowledge influences the competition of organizations crucially. Because of knowledge changing and updating frequently, knowledge engineers need to get knowledge from different sources. Therefore, it is important to integrate relative domain knowledge to support users applying, sharing and reusing knowledge. However, knowledge is an abstract concept which is difficult to define and not easy to integrate. 　　Ontology is an explicit specification of a conceptualization and could share common understanding of the structure of knowledge. It is popular among knowledge engineering domain. Even though ontology could represent knowledge model, it still exits lots of semantic or structure mismatches. Therefore, the core of knowledge integration is ontology integration. 　　An ontology-based knowledge integration mechanism is proposed in this research, including an ontology mapping model and an ontology similarity computing algorithm. This mechanism finds semantic similarity among ontologies and map same concepts between different ontologies automatically to achieve ontology integration. 　　Besides class names and properties of ontology, the “internal-relations” is used as the mapping criteria in this research to get better results of ontology mapping. The integrated ontology achieves knowledge integration, improves system interoperability, and constructs a more robust semantic web to facilitate knowledge searching and reasoning.