CrowdTravel: leveraging heterogeneous crowdsourced data for scenic spot profiling

A multi-aspect tourism information perception method based on multi-source social media data fusion was proposed in order to improve the efficiency of travel knowledge acquisition and provide effective travel assistance for users. A cross-media multi-aspect correlation approach was proposed to connect fragmented travel information after a preprocessing step for the heterogeneous travel related data, which resorted to the “scene-feature characterization” in order to make a multi-aspect characterization. The perception method mined typical travel route from historical travelogues based on the sequential pattern mining, which can be regarded as the recommended route. Connecting cross-media information was based on the similarity between the reviews and the image contexts. Results of experiments over a dataset of eight domestic popular scenic spots, which was collected from Dazhongdianping and Mafengwo, indicate that the approach makes a fine-grained characterization for the scenic spots and the provided travel route can meet different users' needs.