Stories around you: Location-based serendipitous recommendation of news articles

Existing studies in serendipitous recommendation mostly focus on extending the metrics of desired goals such as accuracy, novelty and serendipity with respect to the user preferences. This work aims at serendipity by exploiting the prevailing location (spatial) contexts of the recommendation. For this purpose, we propose a novel spatial context model and a number of recommendation techniques based on the model. A user study on a real news dataset shows that our approach outperforms the baseline distance-based approach and thereby improves the overall user satisfaction with the recommendation result in the absence of the user's personal information.