本研究設計以OWL DL 和SWRL 為基礎的推論系統運用在大學排課問題, 目前推論系統主要作為智慧型代理人特別在自動化、整合和協商等方面。在這些 不同的功能中推論品質的基礎工程在於知識的呈現格式和表達方式。若知識之間 的關係無法呈現或發生衝突,則推論結果將受到影響。語意網是目前新興的技術 可以提供系統之間解譯資訊的能力。語意網主要是以XML、RDF (Resource Description Framework)、RDFS、OWL (Ontology Web Language)等為基礎,透過 這些可將知識進行分類、呈現知識關係和進行推論。 在本研究中主要包含兩個項目,分別是使用OWL 描述知識和使用語意網法 則語言(Semantic Web Rule Language)定義解決問題和發生衝突的解決方案。OWL DL 是OWL 的一個版本,主要利用敘述邏輯表現概念之間的關係,可將實體與 類別分開表示。SWRL 是結合OWL 和RuleML,以本體論為基礎利用語意的方 式定義法則。在實做階段,因為OWL DL 和SWRL 皆無法直接在目前的推論引 擎中使用,本研究採用RACER 連結領域本體論和推論引擎,將本體論中的實例 轉為推論引擎中的事實作為推論基礎,並可在進行推論前針對本體論進行重新分 類和衝突檢驗,確保推論時事實之間沒有矛盾的情形。在法則方面使用XSLT 將 SWRL 轉為推論引擎可接受的格式進行運算。最後我們將此方法運用在大學排課 問題,結果證明本研究的方式可描述複雜的關係,達到知識的分享和再利用,並 結合語意特性的法則,更具彈性的推論系統可解決衝突的狀況。
This study proposes an inference mechanism which is based on SWRL and OWL DL to address ambiguity in course scheduling. Nowadays, the inference mechanism is adopted by intelligent agent especially in automation, integration, negotiation, and etc. Among them, the essential works of inference are formalizing and representing knowledge. The semantic Web is emerging technology as a mean to provide comprehensive capability between machines. It is based on XML, RDF (Resource Description Framework), RDF Schema, OWL (Ontology Web Language), and etc. to construct knowledge taxonomy, relationship, and reasoning mechanism. There are two tracks in this study, using OWL to describe relations among knowledge and using SWRL (Semantic Web Rule Language) to define the actions in conflicts. The OWL DL is one of OWL versions, which performs the concepts of description language to present relationships in a logical form. SWRL is a synergy of combining OWL and RuleML, which enable semantics in define rules. In implementation stage, we utilize RACER to connect domain ontology and an inference engine for generate instances into a repository. Finally, we applied this design in a course scheduling problem. Empirical results suggest that our approach can describe more complex relationship, knowledge sharing, and enhanced semantic capability in rules. It enables flexible inference mechanism to deal with ambiguities in conflict circumstances.