In its most general form, a case base is a source of complex data
stored in specific formats. A number of knowledge and data sources
that comprise a corporate memory could be defined as case bases in
this sense. Case bases could arise from formated records of useful
employee experience and/or expertise. Alternately, Case-Knowledge
Engineers or content experts could design case bases
of relevant experience by populating them with collections of records
in appropriate formats. Certain unstructured databases like text
databases can also be converted to case bases by generating semantic
descriptors characterizing each of its documents. Much of the
work in information extraction and text summarization concentrates on
generating such descriptors[Lehnert et
al.1992]. Given such descriptor
generating capabilities, any set of databases with inter-related data
can be treated as distributed case bases. Another promising
alternative involves creation of metadata[Weibel1995] that is
an informative record attached to a document (structured or
unstructured). A typical metadata record could contain elements like
subject, author, title, object type, relationship to other elements,
coverage etc[Weibel1995]
.
Development of metadata becomes especially feasible if the authors of
a resource or a document could be encouraged to create such a
description. Perhaps some of the most important sources of
distributed case bases available to a corporation beyond its
organizational boundaries are the distributed digital libraries with
mutually related information, like PARTNET[Partnet] on the
Internet. PARTNET is WWW-accessible distributed electro-mechanical
component library developed by the University of Utah's Mechanical
Engineering Department as a resource to connect designers and
engineers with parts suppliers. In this paper, we view a case base as
not just a passive data store but an active one that can reason about its
contents and their appropriateness in the context of a query. Thus,
when we use the term ``agent'' and a case base interchangingly, we
mean that the agent is an active case-based reasoning entity that can
reason about its local case knowledge.