신뢰도 기반의 확률 행렬 분해를 이용한 장소 추천 시스템

Abstract

With the advent of mobile web services, location-based social network services (LBS) became very popular and venue recommendation became an important issue. Many researches have been conducted on using social networks for recommendation, however, the social networks dealt with have only one social object. The most famous LBS, Foursquare, has a social network with multiple social objects. Advancing previous work, we explore each social object’s behaviors to better reflect the real-world recommendation process. In this research, we present trust metrics for users and venues, and propose trust-aware probabilistic matrix factorization technique. For the experiment, we crawl the dataset of the Foursquare and prove that our algorithm outperforms existing methods.