Time-Aware Travel Attraction Recommendation

The increasing number of tourists uploaded photos make it possible to discover attractive locations. Existing travel recommendation models make use of the geo-related information to infer possible locations that tourists may be interested in. However, the temporal information, such as the date and time when the photo was taken, associated with these photos are not taken into account by most of existing works. We advocate that this information give us a chance to discover the best visiting time period for each location. In this paper, we exploit a 3-way tensor to integrate context information for tourists visited locations. Based on this model, we propose a time-aware recommendation approach for travel destinations. In addition, a tensor factorization-based approach by maximizing the ranking performance measure is proposed for predicting the possible temporal-spatial correlations for tourists. The experimental results on the real tourists uploaded photos at Flickr.com show that our model outperforms existing approaches in terms of the prediction precision, ranking performance and diversity.