組合欄位適應性權重與類別核心詞之醫療問答系統研究

Abstract

使用者經驗分享是網路的一種典型行為，醫療問答系統讓使用者不僅能在網路上發表自己的醫療經驗，也能從他人經驗中獲得就醫資訊。透過醫療諮詢資訊網站的資料探勘，醫療知識能更快速的分享與使用，相似醫療案例的分享不僅可以減少醫生重複回答問題的時間，並可協助使用者快速獲取就醫資訊，有效降低知識搜尋所需花費的時間與人力。 本研究針對醫療問答系統提出一個適應性權重相似度計算之方法，結合類別欄位權重與核心詞權重，計算查詢與問答文件之相似度，以提供使用者相關醫療知識。最後，本研究進行實驗評估，比較所提方法在醫療問答檢索之成效。實驗結果顯示，結合權重的方法，確實能有效提高檢索相似醫療案例的成效。

Sharing user experience is a common behavior in the Internet. Medical question-answering systems can be used to help users post their medical experiences and acquire medical information from other users’ experiences. Medical knowledge can be shared and utilized more rapidly through the mining of medical knowledge from the cases posted on medical question-answering websites. With cases sharing on the question-answering websites, medical experts can reduce their effort in answering duplicate questions. Users can also find needed medical knowledge easily without the help of experts. Accordingly, the time and effort spent on searching for knowledge can be reduced effectively This study proposes a method to identify similar medical cases from a question-answering (QA) system. The proposed method combines category core words and field-weight adaption to compute the similarity measures between a user question and a question-answering case stored in the QA system. The identified similar cases are provided to users as relevant medical knowledge. Experimental evaluations are conducted to evaluate the effectiveness of the proposed method. The result shows that our proposed method can improve the effectiveness of identifying similar medical QA cases.