Wild recommendations: Presenting citizens relevant content based on use patterns and context

In this paper we present field experience from the in-the-wild deployment of the recommendation mechanism of CLIO, a novel urban computing system that allows forming and interacting with the collective city memory. Our goal has been to study how users perceive relevant content and in which way the context affects the content they want to explore. We had the opportunity to evaluate for a long-term CLIO in two different cities, in Greece and Finland. The recommendation mechanism of CLIO exploits intelligent techniques in order to present users suitable memories. Intelligence in CLIO is primarily grounded on its ontology-based context model, a reasoning and inference mechanism that support context-awareness and finally a rule-based system producing content recommendations for users based on their random interactions with the system. Our findings during two distinct phases of evaluation shed light on how users consider their profile and preferences, how they perceive relevant content and how the context affects their selections. These results allowed us to develop a recommendation mechanism for the CLIO system, that produces "wild" recommendations for users based on their random interactions with CLIO.