From: pat hayes (phayes@ai.uwf.edu)
Date: 03/23/01
>pat hayes wrote: >[...] > > While I've got your attention on this > > subject... I think I first ran into it > > as 'Assertion of another document' > > in TimBL's semantic web toolbox. > > http://www.w3.org/DesignIssues/Toolbox#Assertion > > > > In my attempts to formalize it, > > > > Ah. I wouldnt try to formalise it in the assertional language >itself, since its really about the process of making assertions >rather than about truth. > >So much for getting your attention on the subject... >I intend to formalize the process of making assertions; >i.e. web protocols. I agree that is worth doing. I was only saying that I wouldnt try to formalize them in the same language you are using them in, is all. That is, it would probably be better to try to keep the language which is asserted, and the language which is saying things about the assertions, separate from each other, at least at first, and once you have that clear, see what happens if you try to smurge them together. Kind of like using mental scaffolding, rather than walking on the bricks while the mortar is wet. I also want to formalize web protocols, but I'm going to try to do it by creating an ontology of web events first, rather than go looking for a new logic. For example, making an assertion is more like an action than a sentence. Which means in turn that no matter how decorated the logical sentences get (tagging and so on) , as long as they are always sentences, they will never get to be assertions. If asserting P is somehow *described* by the sentence 'assert:P', then something still has to assert *that* sentence in order to get something actually done. Sentences by themselves never wake up and get asserted until something asserts them. So there seems to be an essential gap between the actual performance of something - whether its making an assertion or digging a ditch - and a description of that performance in a sentence. The link has to be something like: the description is true iff the action was done (at the time specified, by the agent specified, etc., whatever). Now, that "true" back there is good oldfashioned logical truth in an interpretation, so what we need here isnt a new logic, but a way to ground assertions in the intended world of web actions. Its grounding, not reification, that we need here. Plus of course some language to do the action-describing in, which I suspect will be pretty minimal; I bet we could do it as a DAML+OIL ontology, in fact. >Perhaps not today; but we did talk about getting into >tagged logic or whatever it takes to formalize this stuff, no? >Ah yes... I see JimH has the ball... > >[[[ >tagging and other forms of well-behaved reification. ACTION (Jeff >Heflin, > Jim Hendler): write up SHOE experience. >]]] Yeh, exactly. There are sentences being asserted, and there are tags. One day someone will have to worry about what happens when a tag can tag a tag and one of them says something about itself which contradicts what it says in the very act of doing the saying, or some other vanishing-down-rabbithole type stuff happens; but deal with that later, would be my intuition. I'd be very interested in any other ideas you have in this area, by the way. Pat (Offline most of next week) PS. If I understand tagging, it it needn't involve reification at all. You don't need to *describe* a sentence in order to attach a tag to it, you can just kind of point at it by ostention. And in any case, reification doesnt get you into the sentence/assertion-of-the-sentence distinction, which I think is where one wants to be here. (Maybe I don't really understand tagging, of course.) PPS. Dan, do you agree that the ref document ought to say something about how to make sense of the situation where A imports B and B imports A ? I tried to make it as neutral as possible, but there is no way we can legislate against this happening. --------------------------------------------------------------------- IHMC (850)434 8903 home 40 South Alcaniz St. (850)202 4416 office Pensacola, FL 32501 (850)202 4440 fax phayes@ai.uwf.edu http://www.coginst.uwf.edu/~phayes
This archive was generated by hypermail 2.1.4 : 04/02/02 EST