Cross-Domain Recommendation Method in Tourism

Abstract

In recent years, online travel service has become increasingly popular and effective. Its development has become a hotspot of tourism. Considering the particularity of tourism products, this paper aimed at hotel recommendation service. However, the sparsity of tourism data is an unavoidable problem in recommender systems. In order to solve this problem, we introduce the data from air ticket area, and then builds the cross-domain user profile. In this way, hotel recommendation problem turns to become user profile analysis problem. In addition, we proposed a recommendation method based on transformation matrix, on the one hand, it can solve the cold start problem, on the other hand, users with unsatisfactory results based on user profile recommendation method can be improved. Experiments show that the recommendation method based on cross-domain user portrait is much better than the traditional recommendation method based on popularity or price. Finally, we prove that the recommendation method based on transformation matrix can effectively improve the accuracy of the recommendation method based on user portrait.