Re: ONT Data Models, Ontologies, Logic
Jim,
At 03:40 PM 5/26/2002 -0400, Jim Farrugia wrote:
>Matthew, Adam, et al.,
>
>In response to a recent posting by Matthew (in response to Adam) on the
>SUO list, I am posting this to ONT instead, but feel free to kick
>it back up to SUO if you think it appropriate.
>
>(Note, I see this just crossed John's post to the SUO list...)
>
>
>Matthew says:
>
><little snip>
>
> > MW: An ontology is really overkill for this. Though I should perhaps make
> > clear some distinctions I use related to terms like ontology.
> >
> > - Taxonomy, a dictionary of standard terms and their natural language
> > definitions.
> >
> > - Thesaurus, a taxonomy with subtype/supertype relations.
> >
> > - Data Model, a structure of types and relations against which instances
> > of the types and relations can be stored. Definitions of the types and
> > relations are in natural language. Some constraints are defined, usually
> > in terms of the cardinality of relationships.
> >
> > - Ontology, types, relations, and possibly instances of those together
> > with axioms, usually in First Order Logic, that define the rules that
> > apply to members of the types and relations.
> >
> > Data Models are what are usually developed/used to design a database. The
> > axioms of an ontology are generally of little use/relevance either in the
> > design of the database, or in the population of it by instances.
> >
> > So in general my question for this sort of case would be: Why would you
> > develop an ontology (more work) when a data model will do?
>
><big snip>
>
>1. Would people on this list agree with Matthew's distinctions between
>data models and ontologies?
>
>On the one hand, it does seem that the key distinction is the use of
>axioms, which in turn allow for inferences above and beyond what might
>be stored in a database (based on a data model). On the other hand, one
>often encounters, in doing data modeling, certain constraints, which
>are, in effect, "rules that apply to members of the types and
>relations." Is there a sharp distinction between the kinds of rules
>used to define constraints in data models and the kinds of rules defined
>via axioms in ontologies?
>
>2. And what is the role of logic in defining these rules (for data models
>and for ontologies)? I ask the question broadly, but one particular issue
>I am interested in is the use of implications (if-then statements) to
>express these rules. For the database case especially, such rules seem
>to be expressed as implications. Similarly, I think, for the ontology
>case. But typically the logic used in an ontology goes much further, or
>so it seems to me. For instance, won't any logic used to represent
>an ontology also have pegged to it inference rules, a proof theory,
>and a semantics? Do all these logical extras exist also in, say, a
>relational data model, and is the reason they are not so apparent is
>that the relevant subset of first-order logic is in essence hard-wired
>and implicit in the relational model?
>
>Could someone elaborate on the different ways in which logic is used to
>define the rules both in data models and ontologies, perhaps especially
>focusing on the common logical flavor of if-then rules in both cases?
>
>
>3. Related to the second point above, as well as to the SUO's effort in
>creating an ontology, can one express the kind of ontology we are after
>_without_ using logic? That is, for the kind of computer-crunchable
>ontology that the SUO effort seeks to create, could it exist without
>logic? I suspect that it could, but that such an ontology would be an
>impoverished one.
>
>Two more, if I may.
>
>4. Related to the third point, is a logic-based, machine-processable
>representation of an ontology particularly apt for our purposes, for
>the following reasons?
>
>One could say that ontology is concerned with "what is," the things that
>exist and the relations among
>those things. Logic (plus) model theory is an appropriate way to express
>the notions of an ontology, because the syntax of logic can represent
>things that are, and the model theory of a logic allows us to say "what
>is so." Does this make any sense?
>
>
>5. Finally, related to the points three and four above, and to some recent
>discussion on the SUO list about the language for the SUO (which I think
>have been raised before -- and settled????), I have a question for
>Adam (and John and Chris and whoever else cares to chime in).
>
>Simply put, "Where is the semantics of SUMO?"
>
>Don't clobber me; I think I may have an inkling of the answer, but I need
>some help to make things clearer to me. My inkling is: "Well, the semantics
>of SUMO is given, essentially, via the semantics of KIF (or the version
>that SUMO uses)."
That's correct.
>The semantics of KIF (to be superseded by the semantics of CL) -- well,
>there is where I have a question. On the one hand, there is a model theory
>for KIF at http://logic.stanford.edu/kif/semantics.tex
>
>On the other hand, accordinig to the draft standard at
>http://logic.stanford.edu/kif/dpans.html
>
>"The language has declarative semantics. It is possible to understand the
>meaning of expressions in the language without appeal to an interpreter
>for manipulating those expressions."
>
>I'm guessing that KIF's semantics is defined by the model theory document,
>but how to account for the above mention of a "declarative semantics"?
>
>Thanks,
>
>Jim
Adam Pease
Teknowledge
(650) 424-0500 x571