A Data-driven Approach to Explore Television Viewing in the Household Environment.

The rise of small, IoT-related devices and sensors have enabled us to sense and collect data than ever before. In this study, we walk through our attempt of a data-driven approach in collecting behavioral data on television viewing, an activity thought as passive and habitual. We conducted a 14 day experiment with 13 households in the wild using a data logger installed at each house. Television-related data in IR log data and IPTV packets, and contextual data in Bluetooth signal data and brightness data are collected through the data logger. The data is supplemented by the qualitative situational information that participants provided via in-situ chatbot surveys. Our non-intrusive data logger has enabled behavioral data collection in a natural, comprehensive manner. Detailed television viewing behaviors recorded through IR data logs, volume of viewing sessions, and in-situ chatbot responses show how television viewing is heavily context-dependent than previously thought.