In this research, we design a novel travel recommender sys-
tem to provide on-tour recommendations in a mobile peer-to-peer
environment. Based on the idea that users who visit the same
attractions are more likely to share similar tastes, we proposed
three data exchange methods for users to exchange their ratings
toward visited attractions. We have also implemented a system
that leverages the peer-to-peer data exchange methods for making
recommendation. Through simulated experiments, it was shown
that the proposed data exchange methods that allow data propa-
gating from one peer to another resulted in better recommendation
accuracy than those without data propagation. The operating
regions in which each proposed method achieves the best per-
formance have been identified. From the user study, participants
revealed that, by using the system, they can receive useful infor-
mation at right time and right place, which help them make travel
decisions.
Our work might be extended in several directions. First, some
tourists may not spend the effort in rating attractions they visited.
It would be interesting to design a method that derive ratings from
tourist’s behavior, e.g., the time they stay at an attraction and short
messages they send. Second, it would be more attractive to users
if the system can support various types of social activities, such as
group booking and social gaming. Finally, it would be instructive to
develop an incentive model and security policy to help the adoption
of our service.