|Bernard Omidvar-Tehrani, Sruthi Viswanathan, Frédéric Roulland, Jean-Michel Renders|
|Workshop of State-based User Modelling, co-located with ACM International Conference on Web Search and Data Mining (WSDM) Conference, Houston Texas, USA, 4-7 February, 2020|
Point-of-Interest (POI) recommendation is surfacing in many location-based services. User models are employed in these services to leverage historical check-ins and social links, and enable personalized and socialized POI recommendations. However these models often lack interactivity (incorporating user interactions) and state-awareness. This deficiency aggravates in cold start situations, where nearly no user information (historical check-ins and social graph) is available to generate effective recommendations. In this paper, we propose SAGE, an interactive state-aware POI recommendation system which tackles the aforementioned challenges by exploiting look-alike groups mined in public POI datasets, such as Foursquare and Yelp. SAGE reformulates the problem of POI recommendation as recommending explainable look-alike groups (and their POIs) which are in line with user’s intent. SAGE frames the task of POI recommendation as an exploratory process where users interact with the system, and their interactions impact the way look-alike groups are picked out. Moreover, SAGE defines and employs mindsets which capture the actual state of the user and enforce the semantics of POI interestingness. Our experiments show that SAGE is an effective approach to capture interactivity and contextuality for recommending relevant look-alike groups and their POIs which are oriented towards the user’s mindset.