This study aims to track down representative images and elements of sightseeing attractions by analyzing the photos uploaded on Flickr by Seoul tourists with the image mining technique. For this purpose, we crawled the photos uploaded on Flickr and classified users into residents and tourists; drew 11 region of attractions (RoA) in Seoul by analyzing the spatial density of the photos; classified the photos into 1000 categories and then 14 categories by grouping 1000 categories by utilizing Inception V3 model; analyzed the characteristics of the photo image by RoA. Key findings of this study are that tourists are interested in old palaces, historical monuments, stores, food, etc. and those key elements are distinguished from the major sightseeing attractions in Seoul. More specifically, tourists are more interested in palaces and cultural assets in Jongno and Namsan, food and restaurants in Shinchon, Hongdae, Itaewon, Yeouido, Garosu-gil, and Apgujeong, war monuments or specific artifacts in War Memorial and the National Museum of Korea, facilities, temples, and pictures of cultural properties in Samsung Station, and toyshops in Jamsil. This study is meaningful in three folds: first, it tries to analyze urban image through the photos posted on SNS by tourists. Second, it uses deep learning technique to analyze the photos. Third, it classifies and analyzes the whole photos posted by Seoul tourists while most of other researches focus on only specific objects. However, this study has a limitation because the Inception v3 model which has been used in this research is a pre-trained model created by training the ImageNet data. In future research, it is necessary to classify photo categories according to the purpose of tourism and retrain the model by creating new training data set focusing on elements of Korea.

In this study we aim to analyze the urban image of Seoul that tourists feel through the photos uploaded on Flickr, which is one of Social Network Service (SNS) platforms that people can share Geo-tagged photos. We first categorize the photos uploaded on the site by tourists and then performed the image mining by utilizing Convolutional Neural Network (CNN), which is one of the artificial neural networks with deep learning capability. In this study we are able to find out that tourists are interested in old palaces, historical monuments, stores, food, etc. in which are considered to be the signatured sightseeing elements in Seoul. Those key elements are differentiated from the major sightseeing attractions within Seoul. The purpose of this study is two folds: First, we analyze the image of Seoul by applying the technology of image mining with the photos uploaded on Flickr by tourists. Second, we draw some significant sightseeing factors by region of attraction where tourists prefer to visit within Seoul.

It is necessary to identify the preferences and characteristics of tourists for vitalizing the tourism industry, as the tourism industry is of the fast-growing industries in the economic sector. A trajectory of tourists, a movement of tourists over time, is a very valuable information since it shows tourism characteristics, such as the length of tourists' visit, tourists' preferred attraction, the time of arrival to specific tourist attractions, and the movements in between different tourists' destinations. Earlier studies regarding the movements of tourists were conducted by surveys, or by analyzing the data derived from GPS devices that were handed out to the research areas. However, these approaches using surveys or GPS devices are not only time consuming and requiring a lot of time for the analysis, but also difficult to detect the actual travel patterns. Recently, advances in mobile technologies and multimedia have allowed large amounts of user-generated data, such as travel photos, to be created and shared. This expands the use of such data on tourism industry since it enables to extract and analyze the trajectory of users by using geotagged data that were uploaded on Social Network Services (SNS) (Vu et al. 2015; Zheng et al. 2012).

No abstract