Location-based services (LBSs) are part of our daily lives due to the huge spread of mobile devices. Such services enable us to access relevant and up-to-date information about our current surroundings at any time and everywhere. The adoption of a data-driven semantic layer coexisting with the traditional Web could help further improve LBSs, allowing them to overcome the barriers imposed by closed databases that do not take advantage of the large amount of public data available on the Internet. In this article, we propose a personalized recommender system of points of interest (POIs) located near the user’s current position, which makes use of the gold mine represented by linked open data (LOD). The target user profile is constructed and updated using two differente sources of feedback. The former is obtained by analyzing her activity on social media (i.e., Facebook). The latter is attained by inviting the user to express her interests and preferences as ratings of a sample of selected images representing specific categories of POIs. Experimental tests performed on real users allowed us to verify the good performance in terms of perceived accuracy and normalized discounted cumulative gain (NDCG). Statistical tests also enabled us to verify the significance of all the obtained results.
Sansonetti, G. (2019). Point of interest recommendation based on social and linked open data. PERSONAL AND UBIQUITOUS COMPUTING, 23(2), 199-214 [10.1007/s00779-019-01218-z].