Designing personal assistant software for task management using semantic web technologies and knowledge databases

Abstract

Adoption of social network sites and use of smart phones with number of sensors in them has digitized user's activities in real-time. Smart phone applications such as calendar, email, and notes contain lot of user information and provide a view into user's activities, while sensors such as GPS sensor can be used to passively find information about the user. In addition to this user and device data, these devices have access to the Internet that can be leveraged to build powerful applications. Personal assistant software (smart agent) can be used as an interface to the digital world to make the consumption of this information timely and efficient for the user's specific tasks. Goal of the thesis is to design personal assistant software that understands the semantics of the task, is able to decompose the task into multiple tasks within the context of the user and plan these tasks for the user. It will be designed using semantic web technologies and knowledge databases to understand the relations between the tasks. Agent will be integrated with online web-services to harvest the data available on-line with the data available on the device and help the user to manage his or her tasks. Two use cases are covered in this thesis document to explore automation capabilities and planning capabilities of the agent. Design of the agent using the two use cases helped in the design of sub-modules within the agent system, and also highlighted the requirements on external data and knowledge sources.