應用自發性地理資訊於旅遊網絡之研究

遊客做為觀光旅遊的主體，其旅遊行為在研究中一直受到重視，不同於過去使用問卷或電訪的資料進行研究，自發性地理資訊在Web2.0的背景下，其資料的產製、傳播與應用顛覆了以往傳統的專業取向，自發性地理資訊具有大數據的特性，且屬於使用者生產內容 (User-Generated Content) 的範疇，這些貢獻者 (Volunteer) 生產的資料在時間與空間範疇的分析也突破了以往資料的限制，其標記了時空座標，利用這些資料能探究使用者共同所關注的地點並還原人群的移動軌跡，從而讓自發性地理資訊的資料成為近年地理學新興的研究方向，同時也是研究個人旅遊行為的新媒介。本研究試圖透過遊客所上傳的自發性地理資訊來尋找其所拜訪的熱門景點，並串連旅客的移動路線，進而引入社會網絡分析的指標來探討景點在網絡內的特性以識別景點在整體旅遊環境的重要性，以及景點間的競合，以期應用自發性地理資訊做為新的研究方式，對於臺灣旅遊網絡特性提出較為整體的描述，並透過網絡分析的指標優化網絡，提供未來旅遊規劃與政府相關單位觀光策略的調整與制定。研究從Flickr平台獲取臺灣本島範圍內於2005/1/1~2016/12/31期間32,762位使用者上傳之所有照片共計6,417,506筆，透過P-DBSCAN取得全臺共計838個景點以及166,684筆旅遊路線，根據景點以及旅遊路線建構臺灣的旅遊網絡後，分析網絡的網絡密度、E-I指數、程度中心性、中介中心性、接近中心性、群體中心性以及QAP分析，藉此評估不同景點在旅遊發展中的角色以及區域內的競合，研究結果顯示自發性地理資訊補足了過去研究的缺陷，能夠獲取過去難以偵測的旅客移動之行為，有助於探究更多真實世界的可能性。

關鍵字

自發性地理資訊；社會網絡分析；旅遊網絡

並列摘要

Tourist behavior has always received a great deal of attention in tourism research. For decades, tourism research have been working in the field such as phone interview and questionnaire survey. These surveys are limited by a relatively small sample size. Parallel to the continued use of data from field study, the rapid growth of Internet and the large deployment of mobile devices has led to a massive increase in the volume of records of where people have been and when they were there. These new sources of geo-located information coined as volunteered geographic information (VGI). VGI is also commonly referred to as User-Generated content (UGC), associated with big data and social media, show great potential for geographical research, especially in the field of tourism geography. The aim of this article is to demonstrate the movement of tourist between the most popular attractions. To this end, we used geo-located information contained in Photo-sharing website to identify attractive locations in Taiwan and determine tourist dynamics, then explore the underlying mechanisms of tourist attraction network informed by tourist flows by mean of social network analysis. The trajectories of users that consist of sequences of photos can be used to quantify a network structure where nodes represent the tourist attractions in Taiwan and ties represent the movement of tourists from one attraction to another. Totally, we collected 6,417,506 metadata from 32,762 users over 11 year period, from January 2005 to December 2016. A total of 838 attractions and 166,684 travel routes were obtained through P-DBSCAN. After constructing Taiwan's tourism network based on tourist attractions and tourist routes, we use several indicators such as network density, E-I Index, degree centrality, betweeness centrality, closeness centrality and group centrality to examine the network characteristics of tourist attractions. The results of the study show that VGI complements the flaws of past research and can capture the movement of travelers that were difficult to detect in the past, helping to explore more possibilities for the real world.