AUTOMATED IDENTIFICATION AND SPATIAL DISTRIBUTION OF TOURIST ACTIVITIES FROM ONLINE PHOTOGRAPHS
Abstract (150 Words)
Travel photographs have been identified as an important source of data on tourists’ motion patterns, experience and impressions of the destination. With large volume of imagery data obtainable from the social media, however, the traditional manual processing of the photographs has become a tedious task. Automated classification presents an attractive alternative to manual processing. Yet the viability of this alternative in tourism research has received little attention in the academy. We explored viability of automated identification of attraction locations and leisure activities exercised by the locals and visitors to Lake Texoma, which is one of the largest reservoirs in the United States shared between Texas and Oklahoma. Data mining and spatial statistics methods were used to differentiate locals from visitors, and to identify and classify images to various activities. We found that despite multiple recreational opportunities including well over 50 state- and federally managed parks, space use on Texoma Lake is highly heterogeneous, with recreational activities localized in several hot spots identified as popular attractions. Further, the locals and visitors to the lake tend to differentiate in their popular recreational localities. Eight different leisure activities were identified with respective spatial distributions based on the users’ hashtags. In conclusion, we found the automated recognition of activity type and place from the online photography both reliable and practical.
AUTOMATED IDENTIFICATION AND SPATIAL DISTRIBUTION OF TOURIST ACTIVITIES FROM ONLINE PHOTOGRAPHS
Travel photographs have been identified as an important source of data on tourists’ motion patterns, experience and impressions of the destination. With large volume of imagery data obtainable from the social media, however, the traditional manual processing of the photographs has become a tedious task. Automated classification presents an attractive alternative to manual processing. Yet the viability of this alternative in tourism research has received little attention in the academy. We explored viability of automated identification of attraction locations and leisure activities exercised by the locals and visitors to Lake Texoma, which is one of the largest reservoirs in the United States shared between Texas and Oklahoma. Data mining and spatial statistics methods were used to differentiate locals from visitors, and to identify and classify images to various activities. We found that despite multiple recreational opportunities including well over 50 state- and federally managed parks, space use on Texoma Lake is highly heterogeneous, with recreational activities localized in several hot spots identified as popular attractions. Further, the locals and visitors to the lake tend to differentiate in their popular recreational localities. Eight different leisure activities were identified with respective spatial distributions based on the users’ hashtags. In conclusion, we found the automated recognition of activity type and place from the online photography both reliable and practical.
