On-tour attraction recommendation in a mobile environment

Tourists are often confused about where to go when reaching new and unfamiliar places because there are usually a large number of selections for consideration. In this research, we design an on-tour attraction recommender system that provides personal attraction recommendation for users on the move. In the proposed architecture, each attraction is associated with an attraction server that is responsible for sending ratings to and receiving ratings from peers who visited the attraction. The personal attraction recommender system of a mobile user makes recommendation based on the rating data, which the user indirectly received from other peers via some attraction servers. Four data exchange methods are proposed to allow a target user to exchange rating data with an attraction server. The proposed data exchange methods are compared in a simulated environment. Through experiments, it was shown that the diversity-based method, which mandates a user to carry ratings of other users regardless of their interests, is promising in the case when attraction servers are newly operational and have little historical data. On the other hand, the aggregate method, which relies on the attraction servers to provide recommendation, achieves good performance and incurs least communication overhead in the case when attraction servers have already accumulated a significant amount of historical data.