View article

We apply collaborative recommendation algorithms to photography in order to produce personalized suggestions for locations in the geocoordinate space where mobile users can take photos. We base our work on a collection of 3 million geotagged, publicly-available Flickr.com digital photos on which we applied a series of steps: first, unique locations are identified by discretizing the continuous latitude and longitude geocoordinates into geographic virtual bins; second, implicit feedback is calculated in a user x location matrix using normalized frequency; and third, missing feedback values are imputed through four different algorithms (one memory-based and three model-based). Our results show that two of the model-based algorithms produced the best RMSE and that the RMSE is sensitive to increasing hash bin size.