Multi-Poset-Based Ontologies for
                    Real World Knowledge Discovery

                             Cliff Joslyn
             (featuring Susan Mniszewski, Karin Verspoor,
                   George Papcun, and Michael Wall)

                 Computer and Computational Sciences
                    Los Alamos National Laboratory

                          September 29, 2003
                       Old Dominion University

Ontologies have become an increasingly important concept in modern
knowledge systems, and are cast as one of the foundations of the
Semantic Web. Characterized generally as a "formalization of a domain
of knowledge", they have many different technical realizations, and
ontological concepts have leaked into everything from simple function
typologies to database schemata to object-oriented data structures to
full-up agent-based AI.

One of the very few domains where ontological concepts have escaped
the laboratory and come into their own is computational biology. In
this context, we have been working with ontologies, such as the Gene
Ontology (GO) and the Enzyme Commission (EC) database, which take the
particular mathematical form of large multi-posets, and can be
understood as collections of hierarchical, taxonomic, categorizations
of genes and proteins. Although these are of an intermediate level of
technical complexity, in particular avoiding signficiant logical
capabilities, they are foundational, and of sufficient size and
value to warrant significant attention.

In this talk we will introduce basic concepts in ontologies and
multi-posets, show some of the particulars about the GO and the EC,
and then describe some of our recent research in this area. In
particular, we will discuss pseudo-distance approaches to
categorization in ontologies, methods to regularize mappings among
different ontologies, and the role and value of lexical semantic
approaches in interaction with ontological structures.