Extracting Semantics of Indoor Places based on Context Recognition

Abstract

In this paper we present our work in progress for the dynamic extraction of semantics in a subway station, from smart-phone Inertial Motion Unit (IMU) data. For this, we use machine learning for context recognition and fuzzy set theory for quantification of the uncertainty. Our architecture is client-server, where the clients collect data from sensors, extract features and use classification to identify seven different activities, while these activities are then aggregated on the server with other user activities, from the same regions, to enable the semantic identification of places.