本論文研究目的是要提供一種快速、有效的方式將道路路況資訊中,具有關鍵描述意義的道路資訊擷取出來並轉換成完整的地址資訊,藉以協助在利用地理編碼(Geocoding)的技術將地址轉換成地圖座標前,提供必要的前置處理。路況的資訊在經由人以口語化陳述方式描述時,對於道路名稱的經常會慣用及簡化的方式進行陳述,因此在路況的描述資訊裡經常包含不完整的道路名稱,因此在進行地理編碼時,將會使得服務無法正常解析,又或者是遺漏部份資訊,而使得解析後標示在地圖上的資訊不完整,因此,我們在此研究裡運用正規表式法(Regular Expressions)提出一套處理程序,用來對路況資料中的描述資訊進行擷取及分析,並且使得原先口語化的語句能被處理,進而轉換成地理編碼所能接受使用的地址表示詞彙。In traditional geocoding services, a road name can be mapped to a GPS point by looking it up from a map database. However, most of them do not support partial matching for text descriptions, i.e., to support mapping a sentence containing road names to a GPS point. In this thesis, a geocoding service is designed for extracting road names from text descriptions of accidents based on regular expressions, and the extracted road names can then be mapped to a GPS point from a Map database. Furthermore, a process is proposed to enhance the regular expressions to deal with the problem that road names from the inputs are not contained in the map database from which the regular expressions are constructed.