Intelligent Agents for Retrieving, Filtering, and Managing Information

摘要

Under this grant we developed the first generation technology for rapidly building instructable agents that automate time-consuming or repetitive tasks. In particular, we solved two critical problems towards achieving the larger vision. First, we developed the machine learning technology to rapidly convert semi-structured data into structured data. This allows web sources to be queries as if they are databases, which is necessary for doing any type of additional filtering, processing, or integration on the data. Second, we developed an agent execution system that makes it easy to define agent plans and to efficiently execute those plans. Our research success under this grant is best demonstrated by the fact that we have already published a number of journal articles and numerous conference and workshop papers, filed three pa tents on this work, and licensed the resulting technology and software to a startup company that now has 30 employees.