以時空模型為基礎客製化旅遊行程決策支援系統

Abstract

近年來，個別旅遊(independent travel))逐漸成主要旅遊模式。國人從事國內、國外旅遊分別有超過八成與六成的人以此模式旅遊。本研究的目的是針對個別旅遊者需求，發展客製化旅遊行程決策支援系統。在個別旅遊者遊程決策過程中，有行前決定的固定事項以及行程途中決定的彈性事項兩個階段，需要充分瞭解遊程已使用與未使用的時段以及行前決定行程如何在空間上限制彈性行程的選擇，才能充分利用旅遊時間，但過去研究較少討論時間安排的問題。本研究透過時間地理學的時空稜柱體(space-time prism)方法，結合旅遊路線規劃使用的時間窗概念(time window)，定義潛在旅行區域(potential tour area)以及提出時空窗限制法，作為篩選時空限制下候選參訪點位以及推算最佳參訪時間的功能。時空窗限制法可以協助決策前往目的地前、目的地之間、離開目的地後三種常見的彈性行程安排決策問題。本研究藉由以上方法的定義，完成客製化遊程演算流程以及客製化遊程支援系統。決策流程有六個階段，包含旅遊目的地選擇、遊程內容選擇、必去點位決策、遊程時間分配檢核、遊程內容調整、次要點位篩選。最後，本研究以臺北市為驗證案例，證實本研究的演算法符合傳統最佳路徑演算法的目標，而時空窗限制法最適合在30分鐘以下空閒時間預算下使用。本研究的演算方法能改進傳統以最短路徑為基礎的客製化遊程演算法，節省行程重複演算替代方案的成本，提供局部調整行程功能，以及增加遊程規劃互動過程。

Independent travel is an important tour mode in recent year. Over 80% and 60% people engaged in domestic and foreign travel are independent travelers in Taiwan. In this research, we aimed to develop a travel itinerary customization decision support system for independent travelers. There are usually two stages when independent travelers try to customize a travel itinerary, including a set of decisions made before departure and a set of en route decisions considered to be flexible. However, past research rarely discussed about these issues. Due to this structure of decision process, we must know that the durations of the activities which are decided before departure and how these activities limit the flexibility of en route decisions. We combined space-time prism and the concept of time window to define the potential tour area (PTA). Then, we developed the space-time constraint filter for finding attractions with space-time constraints and calculating how long travelers to stay in these attractions so that we can consider itineraries made before departure as constraints and support travelers to arrange en route decisions. With the definition of PTA and using the space-time constraint filter, we completed a new itinerary customization algorithm and a decision support system. Application to Taipei city, we found that the space-time constraint filter with free time within 30 minutes is more efficient than over 30 minutes. In conclusion, there are three part of improvements of itinerary customization algorithms using the space-time constraint filter that can be more effective than vehicle routing problem algorithms when routing in en route decision problems, adjust part of itinerary without calculate the whole itinerary and have more interactivity when customizing tourism itineraries.