We have given a condensed overview of the UMDL SMS and the agents,
protocols and languages that it encompasses. The system's benefits
are its dynamic extensibility, the ability to write agent wrappers
around legacy database systems (or even other existing web services),
an effective allocation of system resource with little overhead, and a
very open architecture that allows and encourages (via profits) any
third-party to incorporate their agents into our society.
Collectively, these mechanisms automate much of the administration of
a digital library, including organizing information and services using
ontological relationships, selecting, evaluating, and remembering
useful services using machine learning, and deciding how to allocate
(finite) resources to meet the evolving demands of a user community.
These results were achieved via the merging of different technologies
which include: ontology design, market oriented design, and nested
agent modeling/learning. Throughout the paper we mentioned research
issues that still remain open and are under study. Still, we believe
that the SMS has already demonstrated that it can meet most of the our
original goals.
Jose M. Vidal
jmvidal@umich.edu
Tue Sep 30 14:35:40 EDT 1997