I am Professor of Medical Informatics (emeritus) at University of Manchester School of Computer Science.
The purpose of this blog is to provide background, discussion and a place for comments and expansion on issues raised to our recent paper: On Beyond Gruber: “Ontologies” in today’s biomedical information systems and the limits of OWL. (A later strand may discuss more general issues in Health IT.)
My university homepage and professional background are here, but they are unlikely to be be updated for technical reasons. Any further information will be posted here.
As a brief introduction, the word “ontology” has come to be used in so many different ways that even some prominent ontologists admit that there is little agreement on its meaning and perhaps even less agreement about its implications for the architecture of symbolic knowledge representation systems. The various usages of the word have resulted in conflation of philosophical disputes, logical semantics, and technical implementation issues.
The advent of description logics and OWL means that these issues have practical consequences for the design of information systems. OWL’s semantics are often misunderstood or ignored, which limits interoperability, makes the meaning of OWL KBs ambiguous, and has unfortunate consequences for information systems. Indeed, we contend that, for many purposes, OWL’s eclipse of frames and other representations has been a step in the wrong direction.
What can properly be represented in OWL/description logics is limited to statements that can have no exceptions, at least within the scope of a given system – called “indefeasible statements” by logicians. We propose to use the more familiar word “invariants” for the indefeasible statements that form the skeleton of most symbolic knowledge bases as an unambiguous alternative to the word “ontology.” We also suggest further vocabulary and architectures for representing other statements.
For the full argument and vocabulary, see our paper at the link above.
Comments and discussion on the paper and the issues raised is more than welcome.
Further posts will elaborate on these themes.