The Knowledge Modelling Challenge
(A Preamble pre-dating and originally published in the K-Blog itself. September 2001)

A Challenge - Modelling knowledge bumps up against age-old philosophical issues almost before you open the box. Is there a fundamental truth view of the world, and if there is, will we find and agree what it is anytime soon ? Well maybe, but not likely, seem to be pretty reasonable working assumptions; which is hardly a very tight scientific argument, but the challenge is there for any who wish to disagree.

A Warning - So it's important to bear in mind that we're not actually looking for a "Grand Unifying Theory of Everything", even if our practical objective is to achieve something that could be applied to the ubiquitous and generic domain of the entire world wide web, all who may interact with it, and the entire body of human knowledge and artificial intelligence that may represent. I have been concerned with this subject, since a moment I can pinpoint very precisely some 21 years ago in 1980, when I was first struck by an important hidden ambiguity in a pretty insignificant business form, and in particular for the last 5 years concerned with attempting to standardise and implement a generic extended enterprise model in the broadly engineering industry. In that time, I have experienced individuals, myself included, who get drawn towards the fatal attractor at the moment they discover they have a very generic and flexible model on their hands.

The Trick - It is important to remember that when developing your ontology, even when this is a framework, meta-model (or meta-meta-model, or a language etc.) with which to develop an ontology, you are making a choice - deeming which entities may exist. The choice is based on some world view - which may of course be some rationalisation of several other world views - and others with different perspectives and different practical application domains will hold or choose to hold different world views. So a widely applicable generic ontology may come tantalisingly close to being a model to which all others can be deterministically mapped, but there will always be other models to which only incomplete or imperfect mappings will be possible.

The Ologies - Having said that philosophy and meta-physics cannot hold the one true answer to this problem domain, it is of course necessary to appreciate and map between different world-views, which lead to different models or modelling frameworks, and the various limitations and compromises of the different views.

Back to K-BlogHome, WorkinProgress, ResearchResources.