[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: RE: [xtm-wg] [xtm-wg/xtm-iss/xtm-cms] Regarding associations as t opics
There is a very specific formal reason why topic-associations have to be topics (possibly under some view of the topic map) - which I have made to CMS - and I was glad to see Matthew West, with his wider experience than mine in modelling real & seriously complex information domains, agreeing at once from a modelling point of view. The relevant mails are copied below. I am concerned that an apparent implicit requirement from some members that topic-mapping must make immediate sense in English, may be scuppering important formal properties for interrelating (not necessarily merging) topic maps & topic maps, & also for interrelating topic maps and other structured information domains. Ann W. --- HI folks... Here's an interesting preliminary result from my ongoing investigation of OHCO/TM mediated information flow (using Barwise & Seligman's Information Flow logic - see book of same name). 1. In order for a topic map to serve as a "channel" of information about or between one or more clumps of resources, then in this model there needs to be an "infomorphism" from the resource-clump (which I'll call the domain) and the TM. An infomorphism: a) assumes that the types on the domain, and in the TM (=topics) are both classifications (this is why I kept saying "classifier" at Paris...) b) has a function from types in the domain to types (=topics) in the TM c) has a function from particulars in the TM to particulars in the domain. 2. 1c is no problem - if links to resources are effective at all, this happens. 3. 1b is the interesting one. For this to be a *function*, then a type in the domain must, if it evaluates to a topic at all, evaluate to one topic. So, if a TM is to be formally well-conditioned as a carrier of information, a type in the domain which informally we might say maps to two topics, needs to map to an association between them - and this association needs to be itself a topic. So, here is a (preliminary) good formal-modelling reason for topic associations to be first-class topics. Also, whatever syntax and mechanism is decided on for anchoring resources in XTM, the links must be capable of interpretation as *functions* from the TM to the resource domain - though as I said, I really don't see this will be a problem. Enjoy! Ann W. --- Dear Ann, You are right. If you are going to do mapping, associations (whatever links thigns together) need to be first class objects, or else you have no way of saying something like: For each link of type X in model A make a link of type Y in model B, or a link of type X in model A is equivalent to a link of type Y in model B. Regards Matthew -------------------------- eGroups Sponsor -------------------------~-~> Restaurants, Movies, Weather, Traffic & More! Call 1-800-555-TELL. For more info visit: http://click.egroups.com/1/9533/4/_/337252/_/971340544/ ---------------------------------------------------------------------_-> To Post a message, send it to: xtm-wg@eGroups.com To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC