標題: | 以自然語言作為自動路況空間位置標識之研究 The Study of Auto Trafic Information Location Marking—Using Natural Language Descripes Trafic Information |
作者: | 呂欣樺 Hsin-Hua Lu 王晉元 Jin-Yuan Wang 運輸與物流管理學系 |
關鍵字: | 交通路況;語意分析;字串拆解;自然語言處理;Traffic Information;Semantic Analysis;String Parsing;Natural Language Processing |
公開日期: | 2001 |
摘要: | 本研究根據用路人通報路況特性,設計了自然語言自動路況空間位置標識處理流程,透過用路人對於路況的描述,自動將路況描述中空間資訊擷取出來,並透過路況空間定位系統,找出此路況資空間座標資訊,並可將此路況展示在電子地圖上。
首先針對路況自然語言特性,設計了一套路況關鍵詞搜尋流程,增加了路況關鍵詞搜尋之效率與正確性,並且設計關鍵詞鍊結規則表與語意分析規則表,完成關鍵詞鍊結與關鍵此語意分析。並且透過系統的回饋機制,在系統判斷遇到錯誤情形時,可以透過系統操作者修正系統本身缺失,並將系統操作者處理程序紀錄下來,下次如果遇到相同情形時,系統可以自動判斷,達到半自動學習能力。
為測試系統正確性,本研究以警察廣播電台,台北市市區路況、國道1號與國道3號新竹系統交流道以北以及國道2號,實際之路況紀錄資料,作為測試範例,並針對測試結果修正本模式。 This study designs an auto traffic information location marking procedure based on the characteristic of traffic information. This procedure will pick out the desired spatial data contained in the traffic information automatically and determine the associated coordinates so that it could be shown on the electronic map. Because traffic information natural language has it’s own characteristic, this study designs a traffic information key word searching procedure. This procedure increases key word searching efficiency and accuracy. This procedure also proposes a set of key word linkage rules and semantic analysis rules to detect possible errors. If the system finds error, a feedback procedure will be activated to prompt operators to correct errors and update the database. This study uses Public Radio System historical traffic information data as the testing data. The testing results shows that the proposed system is sound and promising. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#NT900423015 http://hdl.handle.net/11536/68681 |
Appears in Collections: | Thesis |