Personalized Instructions for Self-reflective Smart Objects

Due to ever increasing integration of networking and processing capabilities into daily life objects, the Internet of Things (IoT) will affect evermore activities of daily living. These changes require new solutions for improved human-computer interaction. Especially the usage of device ensemble poses entirely new challenges for usability. Currently, we are developing a framework targeting the self-reflection of smart devices in order to counteract such challenges by generating and delivering usage instructions. In this paper, we extend the proposed architecture to consider user preferences during the process of instruction generation and delivery. Besides a user representation on the basis of stereotypes, multiple rule sets are used to identify parameters that are relevant for the whole process. The mapping from users to a matching stereotype relies on the usage of different similarity metrics, among which the Euclidean distance achieves the best results on the basis of a scenario-based evaluation.