On the impact of neighborhood selection strategies for recommender systems in LBSNs

CARLOS RIOS; SILVIA SCHIAFFINO; DANIELA GODOY

Cancún

Conferencia; 15th Mexican International Conference on Artificial Intelligence (MICAI 2016); 2016

SMIA

Location-based social networks (LBSNs) have emerged as a new concept in onlinesocial media, due to the widespread adoption of mobile devices andlocation-based services. LBSNs leverage technologies such as GPS, Web2.0 and smartphones to allow users to share their locations(check-ins), search for places of interest or POIs (Point ofInterest), look for discounts, comment about specific places, connectwith friends and find the ones who are near a specific location. Totake advantage of the information that users share in these networks,Location-based Recommender Systems (LBRSs) generate suggestions basedon the application of different recommendation techniques, beingcollaborative filtering (CF) one of the most traditional ones. Inthis article we analyze different strategies for selecting neighborsin the classic CF approach, considering information contained in theusers' social network, common visits, and place of residence asinfluential factors. The proposed approaches were evaluated usingdata from a popular location based social network, showingimprovements over the classic collaborative filtering approach.p { margin-bottom: 0.1in; line-height: 120%;