| With increasing ubiquity of wearable sensor based devices, smartphones and social applications, rich streams of personal data are being generated with phenomenal volume and variety. The quantified-self movement is increasingly becoming mainstream with easy availability of numerous commercial devices and growing interest in gaining self-knowledge. This presents an opportunity to create a contextual model of a person helpful to gain actionable insights at both individual and society level. But a single device or source may be insufficient to collect data on all aspects of a person's life, thus, multiple data streams need to be combined to detect and store meaningful life events. This poses several challenges related to selection, nature, granularity, reliability and representation.;This work presents a unified framework for aggregating streams of heterogeneous spatio-temporal data, to create, visualize and analyze a chronicle of life events, called Personicle. A formal algorithm for collecting and pre-processing multi-dimensional data has been presented, to abstract high-level context from data streams. A concept lattice based algorithm for life-event recognition has also been explored. A web-based prototype as an application of the framework has been presented, which collects real life data from Google Calendar, Moves application, and Nike+ Fuelband, to create Personicle.;An empirical evaluation of the prototype was done to understand the usability and measure the performance. Eight participants evaluated the web-based prototype for its usability. The user evaluation garnered 75 out of 100 points on the standard System Usability Scale (SUS) scale, thus grading the prototype usability as good. |
