汽車導航系統的便利深受駕駛人好評，它提供路徑規劃和語音路徑指引的功能，讓駕駛人免於迷路的窘境。然而，一般搭公車或捷運等大眾交通工具的行人卻無法享受這種便利的服務，因此我們提出問路代理人系統，針對行人做路徑規劃與路徑指引，提供如汽車導航系統般的服務。本篇論文的重點在於問路代理人系統裡產生路徑指引的部分。
由於問路代理人系統的使用環境與汽車導航系統不同，所以不能使用汽車導航系統的incremental route description，必須使用complete route description來提供完整的資訊給使用者。對於這種路徑描述，我們將從認知與表現兩個層面分析，以得知路徑包含的內容及如何利用這些內容來描述路徑，並根據這些分析結果提出兩種選擇路徑內容的策略。第一種從地標、路口個數、路名及距離等四種特徵之中選出適當的特徵來描述路徑，第二種是當決定以地標來描述時，從數個地標中選擇最適當地標的策略。

當路徑內容選擇出來之後，我們再根據中文語法找出不同路徑描述方式，然後利用Natural Language Generation來產生最後的路徑描述。

Satellite Car Navigation System becomes very popular for car drivers. It provides route plans and route instructions in voice that prevent car drivers from getting lost. However, pedestrians who might want to take bus or MRT cannot enjoy such kind of services. Considering their needs for route advice, we design Route Finding Agent system that provides similar services to the pedestrians. This thesis focuses on the route instruction generation in Route Finding Agent.
The environments for Route Finding Agent are different from the one in Car Navigation System. We use complete route description rather than an incremental one to give complete information to users. As to this kind of route description, we analyze from both cognitive and presentation aspects on how to represent the information. According to these analyses, we propose two strategies to select the route information. The first strategy is to select such characteristic as landmarks, cross counts, road names and distances, to describe a route. When landmark is selected as a characteristic, the second strategy is then applied to choose the most appropriate landmark.

After selecting the route information, we continue to find various ways to describe a route based on Chinese language syntax. Finally, we generate the route descriptions by using Natural Language Generation.

[黃宣範 72] Li, Charles, and Sandra Thompson著，黃宣範譯：漢語語法Mandarin Chinese。台北：文鶴出版社，民國72年。
[運動功能測定] “http://www.cgmh.org.tw/intr/intr2/c3390/cardio2.htm”
[Dale & Mellish 98] Dale, R., and C. Mellish. (1998). Towards evaluation in natural language generation. Pages 555--562 of: Rubio, A., Gallardo, N., Castro, R., & Tejada, A. (eds), Proceedings of the First International Conference on Language Resources and Evaluation.
[De Smedt et al. 95] De Smedt, K. J. M. J., H. Horacek, and M. Zock. (1995). Architectures for natural language generation: problems and perspectives. In G. Adorni and M. Zock, Eds., Trends in Natural Language Generation: An Artificial Intelligence Perspective. Heidelberg, Germany: Springer-Verlag Lecture Notes in AI, No. 1036, pages 17-46.
[Denis 97] Denis M. (1997). "The description of routes: A cognitive approach to the production of spatial discourse", CPC, 16:4, pages 409-458.
[Fraczak & Lapalme 99] Fraczak, L., and G. Lapalme. (1999). A dynamic model of route description.
[Klein 82] Klein, W., (1982). Local deixis in route directions. In R. J. Jarvella and W. Klein, editors, Speech, place, and action, pages 161-182. John Wiley & Sons Ltd.
[Levelt 89] Levelt, W.J.M. (1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press.
[Maaß 94] Maaß, W. (1994). From visual perception to multimodal communication: Incremental route descriptions. AI Review Journal, 8(2-3), 258
[Maaß 95] Maaß, W. (1995). "How spatial information connects visual perception and natural language generation in dynamic environments: Towards a computational model," in Spatial Information Theory: A Theoretical Basis for GIS. Proc. of the Int. Conference COSIT'95, A. U. Frank and W. Kuhn, eds., pages 223-240, (, Semmering, Austria).
[Mann & Thompson 88] Mann, W.C. and Thompson, S.A. (1988). Rhetorical Structure Theory Toward a Functional Theory of Text Organization. Text 8(3)(243-281). Also available as USC/Information Sciences Institute Research Report RR-87-190.
[Marcu 97] Marcu, Daniel. (1997). From local to global coherence: A bottom-up approach to text planning. In Proceedings of Fourteenth National Conference on Artificial Intelligence(AAAI-1997), pages 629-635.
[McKeown 85] Mckeown, Kathleen. (1985). Text Generation. Cambridge University Press, Cambridge.
[Reiter & Dale 00] Reiter, E., and R. Dale. (2000). Building Natural-Language Generation Systems, Cambridge University Press.