Monday, December 12, 2005
A common, widely used, and available ontology
with UDEF (Universal Date Element Framework)
underneath
some points communicated by Cory
comments from
Paul à [296]
Paul,
Sure - we can and should keep it simple and we don't necessarily need logical proofs at this time, but I would also like to make sure we don't naively make it more difficult than it would need to be to utilize formal methods in the future. John Sowa’s paper <need reference here> suggests this should not be a problem[1] .
The basic ideas of context, as I understand them, are quite simple. It provides a way to establish those statements that apply in a situation based on a categorization of those statements and situations by context. It is very easy to see how a procedural or logical formulization could select the appropriate statements for a given set of context [2]. This provides tremendous value in the (realistic) situation of overlapping communities, domains, formalisms and topics. It allows each to exist "in the same space" but segregated by context. [3]
John as suggested "modules", but this sounds very one-dimensional to me, where context should be multi-dimensional. [4] But perhaps this is just a difference in terminology since John's paper provides for multi-dimensional context. That said, even "modules" would help with large-grain distinctions. [5]
"While I do have some ideas about how to model context, I am not sure how to fit it into the formal theories (Breaking each context into an ontology doesn't seem to work as it is too structured). Category theory (Which I don't fully understand) seem to start to give us a way to talk about the relationships between context.
But, I'm not sure we even need that complexity at this point if we can create our set of concepts within a contextual framework, it becomes a separate problem to figure out how to deal with those statements with various formal systems." such as "3D" and "4D".
[1] We are interested in what John has to say about this. The main point is that this separation can occur only if the rational for separating the logic and the concept specification is accepted. This acceptance will be given only under extraordinary conditions, one of them being (1) no work is required to do the separation and (2) that the separation can be reversed with very little work.
There are some fine points about what it might mean to separate logic from concepts, with a possible implication that without the logic the concepts losses some of its meaning. This is an understandable fine point. However, there is also the alternate fine point that any logic imposes restrictions on flexibility and that this flexibility is what human’s use in interpretation of stimulus leading to a concept becoming resident in the mind (of some thing like that).
[2] It is also easy to see, and find examples of cases where, procedural or logical formulization selects statements that are not the complete context, or has misleading information in it. We suggest the separation of a specification of the ontology concepts from any mechanisms for supporting “computational reasoning”, “semantic extraction” or other formal treatment of the concept representations.
[3] There is so much about this statement that is important, in my opinion. Context is naturally a “phenomenological category”. By using this phrase, I mean that something is a context if it has been experienced before. So the “truncation” of actual events (the ontology of a specific event) occurs in ways that are dictated precisely by communities and by repeated action-perception cycles (as discussed a great deal by J J Gibson and the school of ecological psychology that was founded on his work.)
[4] I expect that this is where facets and algebra of toposes comes in. Our concern is that a theory of facets by reflective of some observational science like metabolic reaction pathways. I know that many do not understand why we bring in biology at this point, and this is something that might be discussed further when we examine that part of the SUMO and was merged from the Basic Formal Ontology developed by Barry Smith. The work, the BFO of Smith, has sufficient characteristics and I am encouraged that issues like Edelman’s notion of response degeneracy can be discussed using the ontology concepts extracted from the BFO.
[5] Underneath Sowa’s work there is a specific notion about semantic primitives, and this gives all of his work a degree of dimensional – but the issue is bring forward the degeneracy of viewpoint formation. I need to know the language better here… John?