[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: Re: [topicmaps-comment] Mapping Topics to Cyc: How To Handle Scope?
Thomas B. Passin wrote: > OK, so let us apply this to Murray's earlier test case adn see what we have. > He proposed > > "... a topic "Jimmy Carter" with a base name > "President" (since he could be called that after he was elected) > whose scope is > > scope of titles > scope of English > scope of time "after 1977" > > With "any", we get his title if any of these three scopes is in effect: > > titles, > English, > (OR) "after 1977". > > Agreed? If we are interested in titles, we can find this base name even if > we have not specified "after 1977", as long as we have specified "titles". > That is my understanding of "any". You seem to be dealing with scopes from a querying standpoint. But scopes are used throughout topic maps to combinatorily establish semantic "filters" which affect merging rules, such that incorrect interpretation would lead to incorrect merging behaviours, eg., the topic "Dog" being merged with "George Bush", the former a base name in English, the latter a base name in English (one may have other suitable names for the latter, but that is a different discussion). > That is fine for applying a scope with a single component, but what if we > impose a compound scope consisting of both "titles" and "after 1977"? I was > not thinking of this case earlier. Murray probably was ahead of me and > thinking of this case. Should the baseName string "Jummy Carter" be > considered valid? > > This reminds me of comparisons of node sets in XPath predicates, for which > there are definite rules. For equality, for example, each node in the node > set to the left of the equals sign is tested against each node in the node > set to the right of the equals sign, and if any one node of the one set > matches any one node of the other, the two compare equal (as best I recall). > > Perhaps we should adopt similar rules for Topic Maps. I don't think it would be to anyone's advantage to have different rules for single and multiple components, as it's a likely scenario to (for example) add an additional scope to an existing component while editing, in order to further disambiguate a concept or relation. Having spent all of 2000 on the TopicMaps.Org Authoring Group, thinking about little else than topic maps, many of the use cases are familiar to me. I hesitate to use the term, but I think of this as the "semantic union of scopes", ie., the combination of individual scopes in the linguistic sense. So in the Jimmy Carter example, the scope applied to a topic or topic association is "a personal title, in English, applied after 1977". I think it wrong to assume that this is in any way equivalent to any one of the individual scopes, as before, this might result in merging of *all* topics having base names in English, all topics having temporal extents after 1977, etc. (regardless of meaning). This simply won't do in practice. How to describe this may be up in the air, but from a practical standpoint I guess I don't understand your position. It would seem to create some very strange, erroneous results. Murray ...................................................................... Murray Altheim <http://kmi.open.ac.uk/people/murray/> Knowledge Media Institute The Open University, Milton Keynes, Bucks, MK7 6AA, UK If it wants to be a global power and a player in the Atlantic alliance, Europe has to get back into the business of making war. -- Newsweek Magazine, June 3, 2002
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC