Personicle: Personal Chronicle of Life Events

This paper presents an overview and early results of our project to build a framework for integrating, aggregating, and analyzing heterogeneous personal data streams to build persona and use it to recognize evolving personal situation. Using increasing volumes of heterogeneous personal data streams, we recognize first movements and personal situations for providing individuals actionable information and insights. We demonstrate recognition of high-level life events through simultaneous use of asynchronous observations consisting of continuous GPS and accelerometer measurements. The main contributions of our framework will be to develop an architecture to integrate, store, and analyze data from heterogeneous data streams; detect low level physical activities using various unobtrusive sensors embedded in a mobile phone; and design a hierarchical classifier that identifies high-level life events using location context and physical activity and create a chronicle of life events called Personicle. We plan to collect test data sets and make them available publicly as a resource to other researchers.