A Personalized Scenic Spot Recommendation Algorithm Based on Improved TrustSVD Model

Abstract

Recommendation system is one of the effective ways to solve the problem of information overload. Compared with other recommendation systems, there are data sparsity and cold start problems for tourism recommendation systems, so it should consider more factors. In recent years, researchers have improved and applied various recommendation methods, have proposed many tourism recommendation methods and developed corresponding tourism recommendation systems. In this paper, after analyzing the different problems faced by the tourism recommendation system compared with other information recommendation systems, an improved S-TrustSVD personalized scenic spot recommendation method based on category similarity is proposed that is aiming at the problem for the existing scenic spot recommendation algorithm does not consider users’ personal scoring habits, the viewing effect of scenic spots, as well as the sparsity and cold start of users’ scenic spot check-in data, to analyze and improve the tourism recommendation method integrating users’ social relationship matrix decomposition model TrustSVD. The performance of proposed method with other recommendation methods are verified and compared on the tourism public data set of yelp website. The experimental results show that the proposed method not only improves the recommendation accuracy, but also alleviates the problems of data sparsity and cold start.