本論文提出一個具備語意感知能力的知識本體詞彙庫建構方法，提供語意問答系統在語意流程上的支援，以滿足數位教學環境的需求。透過線上百科做為後端資料，除了建立出同義詞網路，並提出計算語意相似度的方法隔離專有領域的範圍，自動化的流程大幅減少建置知識本體所需的人力成本。 本論文專注在知識本體建置上關係最密切的詞彙資訊，藉由實作來檢驗語意方法的可行性，進而建構出一個領域相關的詞彙網路，這個詞彙網路除了可以做為在語意問答系統中的WordNet替代方案之外，也可作為未來設計更有彈性的知識架構的參考模型。 In this thesis, we proposed an automated ontological creation process, and focus on domain terminology’s relationship. By analyzing web pages on Wikipedia, an online encyclopedia, we constructed not only a dictionary but a semantically related words’ network. In order to fulfill the semantics’ need, we also proposed an algorithm to calculate the semantic distance between terms. The entire process can greatly reduce the human resource requirement in building a traditional domain Ontology. This system also provides a better support for domain-specific QA systems than a general purpose lexical dictionary in its flexibility and further offloads the semantic process burdened on QA systems.
