以自然語言為基礎的交通路網通阻定位系統之研究

臺灣的颱風、地震災害頻繁，搶救災害的同時，交通路網的通阻狀況扮演相當重要之角色。目前國內的交通路網通阻系統多半已與地理資訊系統結合，利用網格格式或向量格式的電子地圖作為系統展示底圖，通阻資料的建置則有兩種，一是人工數化方式（使用者用肉眼判斷所欲建置之通阻點資料位置，以滑鼠於螢幕上數化），另一是座標定位方式（取得通阻點的經緯度座標或TM二度分帶座標）。前者主要缺點在於數化點位置不易準確，誤差較大且耗費時間較長；後者主要缺點則在於災害發生時座標資料不易取得。考量救災時間急迫，本研究因應現有通阻資料多以文字方式呈現（如：交叉路口、著名地標、某道路幾公里處等），設計建構一個以自然語言概念為基礎，具備判釋文字資訊並自動定位於路網功能之交通路網通阻資訊系統。分析通阻描述資訊的文字結構，並歸納出幾個主要的描述類別，再結合自然語言關鍵字對應法研發系統中各描述類別之自動定位功能。以臺北市文山區作為研究示範區，利用道路與地標圖層進行空間定位分析，完成系統主要功能：輸入一段具備空間意涵的文字資訊即可自動定位於電子地圖上，並存成一筆新的點資料，作為災時決策之參考。

關鍵字

通阻定位；語意分析；自然語言；動態路段；地理資訊系統

並列摘要

Numerous natural disasters have attacked Taiwan and caused severe damage and loss in the past and it is not surprising that mother nature will strike again. However a more effective response can significantly reduce the damage and loss. Blockage information of the transportation network is key to an effective response. Currently officials have combined GIS technology with their blockage information system by using digital maps, either in grid format or vector format, as background for display. There are two different ways used to generate blockage records. One is to have an operator identify the blocked location from a digital map and then digitize it manually. The other is to automatically identify the point on the digital map by key in the coordinate of the blocked location. Both ways have their own disadvantages. The first one lacks accuracy and it is time consuming while the second one is that it is not easy to get the coordinate values from the current reporting system. Normally a block situation would be reported by verbal descriptions such as a road intersection or mileage of a certain route, by classifying different types of reporting descriptions and analyzing the construction of such description. This research is aimed to study the possibility of combining a natural-language concept with GIS technology to establish a system that can parse the verbal description and identify the reported location automatically. A prototype of the system has already been established. The system can identify the location on the digital map automatically by keying a geographical description and save it as a record. A case study of Wen-San district in the Taipei basin has been done and demonstrated in this paper.

並列關鍵字

blocked location ； semantic analysis ； natural language ； dynamic segmentation ； geographic information system

參考文獻

Allen, J. F.(1987).Natural Language Understanding..Menlo Park, Calif.:Benjamin/Cummings.

Google Scholar

Brill, E.,Mooney, R. J.(1997).An overview of empirical natural language processing.(AI Magazine).

Google Scholar

Cecil, W. H. G.(1996).Location Referencing for ITS.Oak Ridge National Laboratory.

Google Scholar

Cecil, W. H. G.,David, S.,Stephen R. G.(1996).Location Reference Message Specification Preliminary Description.U. S. A:Federal Highway Administration Office of Safety.