文字探勘(Text mining)結合資料探勘、自然語言處理與資訊檢索技術，使大量不具結構的文字資訊能經由電腦自動的分析歸納。其中一個重要的主題「文件摘要(text summarization)」，經由電腦自動依照其文件內容，選取出重要的句子組合形成文件摘要，從而減少學習者閱讀時間並且增加效率。 本研究提出ㄧ種以本體論為基的文件段落擷取方法，以期能有效的協助網路學習者在網路文件上獲取正確且快速的相關知識。本實驗結果也發現，以多個關鍵字詞所擷取的文章段落比單一字詞擷取更能符合學習者的語意。藉由領域本體論與自然語言處理的結合，把使用者所輸入的問題進行分析，找出文章中符合語意的段落回應給使用者，促進知識的分享與再利用。 Text mining combines data mining, information retrieval and NLP (Nature Language Processing) technologies.Automatically text summarization is a problem which is using computer technologies to give an user quick and useful document's knowledge. In our research, Ontology can help us to define a word concept and relationship. We propose a multi-keyword method to judge a document's paragraph whether fit user's semantic. When a user input a query, we should according to user's semantic to find the suitable document's paragraph, and representation it to the users. It can be promoted knowledge reusability and sharability. And our experiments results finds that using multi-keywords extract document's paragraph can suitable user's query.
