POI Recommendation through Content Preference Location Aware Collaborative Filtering

Abstract

The focus of this article is on recommending point (places) of interest (POIs) to a user based on the user's present location and previous check-in behaviors. To solve this issue, we separate user preferences for POI content from user preferences for POIs themselves.  The final POI suggestion is obtained by combining the projected content rating with the predicted location rating of the POI. We test this strategy with data from Yelp and Foursquare. This technique outperforms the competition and works well in "new city" situations if the user has not rated any of the POIs in the current region.