Where Should I Go? City Recommendation Based on User Communities

Abstract

Recommender systems play a key role in the decision making process of users in Web systems. In tourism, it is widely used to recommend hotels, tourist attractions, accommodations, etc. In this paper, we present a personalized neighborhood-based method to recommend cities. This is a fundamental problem whose solution support other tourism recommendations. Our recommendation approach takes into account information of two different layers, namely, an upper layer composed by cities and a lower layer composed by attractions of each city. It consists of first building a social network among users, where the edges are weighted by the similarity of interests between pairs of users, and then using this network as a component of a collaborative filtering strategy to recommend cities. We evaluate our method using a large dataset collected from Trip Advisor. Our experimental results show that our approach, despite being simple, outperforms the precision achieved by a state-of-the-art baseline approach for implicit feedback (WRMF), which exploits only the overall popularity of cities. We also show that the use of a secondary layer (attraction) contributes to improve the effectiveness of our approach.