Principles to Design Smart Physical Objects as Adaptive Recommenders

Recommenders have proven to be useful means to support people in their activities and in making decisions. They evolved from online recommenders to context-aware and ubiquitous recommenders. Moving forward along this line, this paper introduces the new emerging class of smart physical recommenders: context-aware recommender systems that are embedded into physical everyday objects. This paper describes the features of these systems and presents a conceptual model to design them, by analyzing a number of issues that have to be addressed by a designer and discussing the consequences of different design choices with their impact on the smartness of the designed object. The model is structured in a number of layers corresponding to different conceptual design phases in which different requirements are analyzed. The contribution of this paper is to discuss and provide design guidelines for a new rising class of recommenders that combine the features of intelligent agents, cyber-physical objects, and recommender-support systems. The description of the model is complemented by an exemplary analysis of its application.