Intelligent Personal Agent
November 1, 2003 - May 1, 2003
Objective. The objective is to develop a personal agent that can provide and maintain the contextual information about a user. Acting as a record-keeper, this agent tracks a user's context by acquiring information from external sources (e.g., the context brokers in different smart spaces). Acting as an information provider, this agent shares the updated context information about a user with systems that attempt to provide services to the user.We assume the users are willing to share certain private information with the personal agents that serve them (e.g., the users are willing to share their Outlook calendars, contact lists, work locations, living behaviors and patterns with their agents). We also assume the users trust the decisions made by their personal agents to share information with other systems.
Design Requirements. The personal agent should be designed and implemented as a FIPA-compliant agent (i.e., using FIPA communicative acts to share information with other systems, obeying the FIPA agent life cycle defined by the FIPA specifications). Information exchanged between the personal agent and other systems, including the context brokers, should be represented using the Web Ontology Language OWL, if possible. Information that describes the context of a user should be represented in according to the CoBrA ontology.
Deliverables. (i) Documentations of the types of contextual information that a personal agent can provide (e.g., the location of a user, whether a user is in a meeting or not, whether a user has arrived at work). (ii) Documentations of the types of external sources from which the personal agent acquires and builds a model of the user's context. These documentations should also identify the necessary technologies required to realize the acquisition of different contextual information. (iii) A prototype implementation of a personal agent that can interoperate with the context broker in the EasyMeeting system (i.e., the agent can acquire certain contextual information from the broker, and it can provide the context broker with certain information about the user).
Contact Information More information on CoBrA can be found at http://cobra.umbc.edu. If you are interested in this project, please contact Harry Chen and Anubhav Sonthalia