以領域本體為基礎跨語言糖尿病資訊問答系統之研究

The study of Cross-Languages Diabetes Information Question Answering System Based on Domain Ontology

吳威翰(Wei-Han Wu)

指導教授 ： 陳靖國

朝陽科技大學/資訊學院/資訊管理系/碩士(2013年)

摘要

近年來糖尿病病患逐年增加，且隨著網路技術迅速發展，使用者可透過瀏覽器以及搜尋引擎來查詢相關資訊，但目前搜尋引擎主要遇到兩個障礙(1)語言障礙；(2)搜尋障礙。對於搜尋障礙方面，由於在資訊爆炸的時代，搜尋引擎所查詢回來的結果種類太多，內容繁雜。使用者必須要逐一的瀏覽內容，才能確認所搜尋回來的結果是否正確，是不是使用者本身要的內容資訊，這是一種費時費工的工作。因此問答系統技術越來越被受到重視，問答系統的目的是使用者提出任意問題，系統給予一個正確的答案。對於語言障礙方面，使用者使用搜尋引擎的另一個問題使用者可透過搜尋引擎來查詢相關資訊，但獲得的資訊只限於和查詢關鍵字相同的語言。因此造成搜尋語言障礙的問題，並導致使用者可搜尋的資訊量減少。本論文的主要目的是要結合特定領域的知識本體、問答系統和跨語言資訊檢索技術，開發一個跨語言糖尿病資訊問答系統，提供使用者輸入查詢問句，並透過自然語言處理程序了解使用者想要什麼樣的資訊，再藉由多語言本體提供合適正確的答案。實驗結果證明我們所提出方法具有70％以上的正確率在回答糖尿病知識問題上。

關鍵字

糖尿病 ； 跨語言資訊檢索 ； SPARQL ； 問答系統 ； 知識本體

並列摘要

In recent years, diabetic patients are increasing quickly. And with the rapid development of network technology, users can query information through the browser and search engine. However, there are two main barriers for search engines. The first one is language barrier and the second one is search barrier. In search barrier, the query results have many species and the contents are too complex that users must look for the contents of web pages in order to confirm the search results are correct and the results are users needed which waste much time and work. Therefore, a technology of question answering system has been more attention. A question answering system provides relevant information to users for correct answers questions. On the other hand, users use search engines exist the other problem that the access information is limited to the same keywords and query languages. This is a language barrier which leads to information reduction. The main purpose of this study combined domain-specific of the ontology for the question answering system with cross-language information retrieval technology to develop a diabetes information cross- language question answering system. The system offers users to inputting query questions and to understand what information users want through the natural language processing based on multi-language ontology to provide appropriate and correct answer. The primary experiments prove that our system has more than 70% accuracy for on line language query and answer for diabetic information.

並列關鍵字

Diabetic mellitus ； Ontology ； Cross-Languages Information Retrieval ； SPARQL ； Question Answering System

參考文獻

[10] 行政院衛生署統計室

[2] 陸韋銘(2009)，以本體論為基礎運用模糊正規化與SPARQL查詢語言之研究，碩士論文，朝陽科技大學資訊管理系碩士論文，台中。

[4] 陳慶龍(2007)，可供語意搜尋之部分知識建構，碩士論文，國立成功大學工程科學系，臺南。

[14] S. J. Athenikos and H. Han (2010), "Biomedical question answering: A survey," Computer Methods and Program in Biomedicine, Vol. 99, No. 1, pp. 1-24.

[19] S. Ferrandez, A. Toral, O. Ferrandez, A. Ferrandez and Rafael Munoz (2009), "Exploiting Wikipedia and EuroWordNet to solve Cross-Lingual Question Answering," Information Sciences, Vol. 179, No. 20, pp. 3473-2488.