Point-Of-Interest Recommender System for Social Groups

Abstract

This paper presents a signature based point of interest (POI) group recommendation system which provides personalized recommendations of places, such as restaurants, for mobile social groups. The growth of location-based social networking (LBSN) like Foursquare in recent years allowed users to explore POI’s easily. While the POI recommendation for individual users has been studied greatly, there has not been much work done when it comes to recommendation for group events. The recommendation for groups is different to that of single user as the requirements vary in both cases. In this work, we identify the places which are used for group events and then build signatures based on group size, location, time and other parameters. Then, for each new group we build a similar signature and recommend places based on similarity with the calculated signatures. The experimental results demonstrate how the algorithm exploits the structure of the group and gives prediction based on the group rather than the approach of giving predictions to single user.