ConferenceCall 2004 11 04

[Professor Christopher Menzel]    (3XR)

Abstract    (3XS)

The rapid growth of knowledge engineering has seen a concomitant proliferation in the number of knowledge representation languages -- from expressively rather weak, graphically-oriented database modeling languages like ER and IDEF1 to the sophisticated outgrowths of is_a hierarchies known description logics to full first-order logic as exemplified in frameworks like KIF and John Sowa's Conceptual Graphs.    (3XT)

The growth of the World Wide Web has made it imperative that we be able to *integrate* distributed knowledge bases -- medical researchers can benefit from research compiled in laboratories thousands of miles away; businesses wishing to collaborate on a new manufacturing enterprise need to share their planning and production models; and so on. Typically, however, the information in distinct knowledge bases will be expressed in frameworks that can differ wildly in both form and expressive power.    (3XU)

Given the obvious inefficiency of attempting to integrate representation languages piecemeal, a natural approach to a solution to the problem of integration is the idea of an "interlingua": an expressive representation language that could serve as a sort of neutral framework for facilitating translation and integration between different representation languages. KIF was in fact initially envisioned as such a framework, though it seems to have become rather more popular as simply a way to write first-order ontologies at a computer keyboard. There is in any case an obvious problem with using KIF as an interlingua, namely, it too is simply Yet Another Representational Framework; a good one, to be sure, but one with its own quirks and liabilities.    (3XV)

SCL -- Simple Common Logic -- is an attempt to renew the idea of an overarching framework to facilitate integration at one higher level of abstraction: by specifying, in as general a way as possible, a flexible abstract standard that many different representation languages can satisfy. By publishing an abstract specification of one's language relative to the SCL standard, one thereby makes the representational character and expressive power of the language explicit, and information expressed in the language comprehensible to users of other conformant representation languages.    (3XW)

In this (very informal!) presentation I will discuss SCL and also use it as a springboard for discussion of a number of related topics, including current popular KR frameworks -- notably frameworks arising out of work on the Semantic Web like RDF -- and the expressive differences between them.    (3XX)

I received my PhD in philosophy from the University of Notre Dame where I wrote a dissertation on the philosophy of mathematics. I was then a postdoctoral fellow at Stanford University at the Center for the Study of Language and Information (http://csli.stanford.edu), where I first became interested in knowledge representation. I have been a visiting research scientist at the Commonwealth Scientific and Industrial Research Organization (CSIRO) in Melbourne (http://www.csiro.au) and am currently visiting the Institute for Formal Ontology in Medical Information Science (IFOMIS) at the University of Saarland in Saarbrücken, Germany (http://ifomis.org).    (3Y3)

My research interests reflect my initial background in "pure" metaphysics, formal ontology, and logic and my subsequent introduction to "applied" philosophy in the form of knowledge representation. On the pure side, my research focuses on understanding the logic of the so-called alethic modalities of necessity and possibility. I am particularly concerned with developing a very robust logic of modality that preserves our strongest modal intuitions but avoids ontological commitments that, with my philosopher's hat on, I find objectionable. On the applied side (where I tend to run roughshod over my philosophical scruples), I attempt to apply the tools and techniques of formal ontology and mathematical logic representing and managing information in engineering and manufacturing systems. I have spent a lot of time working on problems of representing dynamic information, especially in the context of the early development of the PSL project (http://www.mel.nist.gov/psl) and a couple of papers with Michael Grüninger. More recently I have worked with Pat Hayes and others to develop a "(Simplified) Common Logic (SCL)", a framework for characterizing logical languages and their interpretations in wholly abstract terms (http://cl.tamu.edu). The hope is that thereby we can establish a clear and rigorous standard for writing ontologies in a logical language on the web that does not legislate any particular concrete that language should take. One need only express how the abstract constraints specified by SCL are satisfied by the language one is using. In so doing, one thereby ensures the proper interpretation of one's language, at the logical level, at least, and facilitates the ability to translation of ontologies in that language into other knowledge representation languages.    (3Y4)