LSVP: A Visual Based Deep Neural Direction Learning Model for Point-of-Interest Recommendation on Sparse Check-in Data

Abstract Recently accumulated massive amounts of geo-tagged photos provide an excellent opportunity to understand human behaviors and can be used for personalized POI recommendation. However, no existing work has considered both the visual contents in these photos and the sequential patterns of users’ check-ins for POI recommendation. To this end, in this paper, we propose an attentional network named LSVP for POI recommendation, which adaptively considers the joint effects of users’ long-term, short-term and visual preferences. Specifically, we first extract visual preferences from photos, then extract long-term and short-term preferences from check-in sequences. At last, an adaptive attention mechanism is used to balance all the extracted users’ preferences. Experimental results on two real-world datasets collected show that LSPV provides significantly superior performances compared to other state-of-the-art POI recommendation models in terms of accuracy.