隨著電腦與網際網路的蓬勃發展，利用網路搜尋資料已漸漸成為時代趨勢。在目前通用的搜尋方法中，都以單一或多個關鍵字做為搜尋重點，但往往都不能按照自己意圖的描述方式來做搜尋，所以會花費更多時間來找尋不符合意圖的資料。因此，要如何去達到符合自己意的的描述方式來找答案，是目前學界所追求的進一步研究目標。 本研究發展一解析系統，主要是來分析中文問句的意圖，此系統包含三個步驟，第一以中研院CKIP系統做斷詞及詞性標記，且以5W1H系統化的歸納方法來搭配有限自動機(Finite Automata)的比對手段來解析出意圖型態以及關鍵字，第二以關鍵字來搭配領域本體論(Ontology)做擴展成語意網，第三將解析後所對應的5W1H對應字與語意網做結合。在實驗部分，我們會傳統單一關鍵字與本研究所提出的關聯對應字來做結合，將結合後的字串於網路搜尋引擎中做搜尋處理。其結果部分，我們可以改善其搜尋的準確率約5%至34%效能及可縮短搜尋時間。 With computer and the vigorous development of the Internet, use the Internet to search for information has been gradually become the trend of the times. In the current search methods, a description of the query search will often be the key, but often do not accord the user's intentions, then accuracy was challenged. Questions should be to understand how the user's intention is to pursue further research objectives. In this paper, this research mainly makes analysis of the Chinese question sentence, it contains three steps, the first step is break sentence segmentation and Part-Of-Speech by Sinica CKIP (Chinese Knowledge Information processing) system, and find intention type and keyword by 5W1H induction and the Finite Automata method, the second step expands the semantic net by key words matching domain Ontology, the third step the 5W1H correspondence word and the semantic net do the association to the search engine does the search. This research result showed we proposed joins the 5W1H correspondence word method to be possible to improve its search the precision , can helps user to find intention answer by network.
