Finding Top-$k$ Places for Group Social Activities

Abstract

Geo-social network applications utilize check-in information to suggest places for social activities. This paper focuses on recommending points of interest (POIs) to groups of users based on the current location of users and the popularity and suitability of the POIs from history. To address the problem, we propose a new type of query, namely, group-based geo-social top-k places (${\textsf {G}}k{\textsf {P}}$) query, which takes spatial proximity and social fitness into consideration. This is among the first attempts, and we present the preliminary results. In particular, we investigate the problem formulation, especially the modeling of spatial proximity and social fitness. Two baseline algorithms, distance-driven and relevance-driven, respectively, are conceived. Initial empirical results confirm that ${\textsf {G}}k{\textsf {P}}$ queries meet the needs of potential applications, and the proposed algorithms are sufficient to handle ${\textsf {G}}k{\textsf {P}}$ queries.