Cycorp – Dimensions of Context Space

Cycorp – Dimensions of Context Space. Reference to Doug Lenat’s paper (via Seth). If one accepts “context” as the space in which information represented by relationships is organised and characterised, then this Cyc stuff has a lot in common with the EPISTLE work I’ve mentioned before. Also like EPISTLE Atomic Templates and Seth’s Mentographs, includes the idea that information comes as “atoms” which may be indivisible in terms of their communication, but have internal structure which together with context defines their semantics. I see a total convergence of thoughts here, in that the description of “information” is reduced in all cases to characterisation of relationships and the directed pattern of their linking, even if I’m uncomfortable with the choice of “context” to describe this set. (Lenat’s context – Figure 1 – in fact seems to represent what I alluded to as the communication chain in my manifesto, originally brought to my attention by Guarino.)

Lenat’s paper goes further, in that it suggests (defines) 12 dimensions for characterising these “contexts”. The basic spatio-temporal-extent (EPISTLE) idea is there as both points and periods in both time and n-Dimensional space (four of the dimensions). Most of the remaining 8 are very much around human intent, which is very encouraging, but I see no convincing argument offered as to why this particular breakdown is significant, though examples given demonstrate it’s utility. Pleased also to see “granularity” as an explicit dimension – my current spin is one of “fractality” – something like necessary levels of granularity linked to complexity of the context.

I ought to explain why I don’t feel bound to respond scientifically to the 12 dimensional breakdown from the perspective of attempting to challenge any specific inadequacies. There may in fact not be any, but that is not my point. Taxonomies / ontologies also can be very useful and flexible, so much so that many people would still not question significant limits to their utility for organising information / knowledge. I am more concerned at present to establish if there are any fundamental bases on which such ontologies (even ontologies of context – because that’s all they are) should be constructed – before re-addressing any pragmatic limits to implementation.

Lenat’s 12 dimensions do indeed however look like the shape of things to be expected. No argument there.

[Note to self – must get into the habit of creating little discusions like the above as “quickies” – a la Jorn – so that they take up less space in the Blog itself.]

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.