Towards Sensing and Sharing Auditory Context Information Using Wearable Device

Abstract

Data-driven information services using wearable devices have attracted attention in the areas of healthcare, medical care, and educational services. In the services, the users’ daily behaviors are modeled with the sensing data of the physical statuses of individual users, e.g., body movements, heart rates, etc. However, to understand human behaviors more deeply, it is also important to know the context information of the users, such as the surrounding environment and participating activities. In this paper, we describe extracting auditory context information from ambient sound data sensed by smart watches. First, we describe a prototype of our wearable ambient sound sensing system by using smart watches. Then, we describe an analysis of the sound data sensed by the system. We formalize the context extraction process as unsupervised segmentation of multi-dimensional time-series data and apply non-negative matrix factorization (NMF) and k-means clustering to the segmentation at the first step of the study. We confirm that the periods segmented by the analysis roughly correspond to actual contexts.