Agent Rules for E-Commerce: Contracting and Trust on the Semantic Web
Speaker: Benjamin Grosof , MIT Sloan School
Date: March 15 2001
I give an overview of my current work: on knowledge representation (KR) for communication of rules, beliefs, and trust between multiple parties ("agents") on the Web, applied to contracting and negotiation. The KR focus is largely on rules, using several of my novel extensions of declarative logic programs. One extension is to enable XML-based semantic inter-operability between heterogeneous commercially important families of rule and database systems, and thereby between the agents that use such systems. A second extension, based on a theoretical advance, is to enable prioritized conflict handling in an expressive yet compuationally tractable fashion -- this represents one of the first *practical* expressive generalizations in over two decades of research in non-monotonic reasoning. A third extension is to enable a disciplined and inter-operable form of dynamic procedural attachments.
I discuss the requirements to automate Web (agent) contracting, negotations, and auctions. I then show how the approach enables contracts, bids and auctions to be specified modularly, modified dynamically during negotiations, and executed automatically after import into multiple agents' heterogeneous e-business processes. It also enables policies for authorization and trust, including with delegation, to be specified likewise and to have provable guarantees about the behavior of their implementation. We have prototyped an earlier version of this approach as IBM CommonRules (available on IBM AlphaWorks), and piloted that in several application settings, including in EECOMS, a just-completed $29M 3-year NIST ATP project on supply chain collaboration in manufacturing, by an industry-university consortium that included IBM, Baan, Boeing, and TRW. Current efforts include grants under the DARPA Agent Markup Language program and the Center for E-Business@MIT Vision Fund, as well as a budding industry standards effort called Rule Markup Language.
See other events that are part of AI Colloquium Series Spring 2001
See other events happening in March 2001