Enabling an ecosystem of personal behavioral data

Almost every computational system a person interacts with keeps a detailed log of that person's behavior. The possibility of this data promises a breadth of new service opportunities for improving people's lives through deep personalization, tools to manage aspects of their personal wellbeing, and services that support identity construction. However, the way that this data is collected and managed today introduces several challenges that severely limit the utility of this rich data. This thesis maps out a computational ecosystem for personal behavioral data through the design, implementation, and evaluation of Phenom, a web service that factors out common activities in making inferences from personal behavioral data. The primary benefits of Phenom include: a structured process for aggregating and representing user data; support for developing models based on personal behavioral data; and a unified API for accessing inferences made by models within Phenom. To evaluate Phenom for ease of use and versatility, an external set of developers will create example applications with it.