- From: pat hayes <phayes@ai.uwf.edu>
- Date: Sun, 29 Sep 2002 17:08:36 -0500
- To: las@olin.edu
- Cc: www-webont-wg@w3.org
Lynn: I actually largely agree with you. But you are arguing here from a position which is WAY out in left field (if you will forgive the use of the standard metaphor for a moment) and I was arguing against a position which is so right-field as to be unworkable even as a quick dirty compromise first step. So it seems to me that in this particular discussion we are in fact arguing on the same side. >pat hayes wrote: > >> (Webont members: I am BCCing this to webont as well as the DAML JC as >> I think the issue is relevant to our forthcoming F2F discussions. >> Hopefully BCC will avoid accidental cross-posting. -Pat) > >Unfortunately, I can't make the F2F (though I'm still hoping there's some >opportunity for remote participation) and so will miss that part of the >conversation. > >> Ian writes: >> >E.g., in a database context, >> >querying could be viewed as saying something like "assuming that what >> >I have here are all the tuples/elements in the universe, does it >> >follow that ...". This does not mean that I am asserting that what I >> >have here really are all the tuples/elements in *the* universe. >> > >> >Taking your argument to its logical conclusion would mean rejecting >> >pretty much any assertion you could make as it is bound to be false >> >w.r.t. something somewhere on the web. > >> Pat responds: >> I strongly disagree with that. Where on the web is going to be false >> that there are infinitely many integers? Where is it going to be >> false that Paris is the capital of France? Etc. Surely the point of >> having information publicly available on the SW is so that agents can >> assume that it is all just plain true, rather than true only in some >> unknown context. > >Well, to a C programmer int is restricted to 32 bits and in certain parts of >Texas Paris is the town next door. True but beside the point. Where on the web is going to be true that all integers are less than 2|32? Well, maybe inside C code, in a very warped sense; but nowhere else. Where on the web is it true that all C numerals are less than 32 bits long? Everywhere. And if the guy living near Paris, Texas publishes some RDF which use 'paris' to refer to a town in Texas, relying on his context for us to know what he means, then he has no right to complain if some piece of software in Europe misunderstands him and his packages get delivered to the wrong place. Tough shit: the web is bigger than Texas. But you know, I think people understand all this very clearly already, and already know the need to decontextualise what they say if they want to be widely understood. I bet that the guy in Texas wouldnt expect to use Orbitz fly him to Paris, Texas rather than Paris, France. I don't think that this whole issue is going to be a real problem in practice. Take a look at the actual data that is out there already, such as Guha's TAP KB. How contextual is it that Kevin Kline is an actor? On the web, not at all. The web is *already* globalized and decontextualized, to a large extent. > I realize that one response to this is that >what a C programmer means by an int is not the thing you meant pjh:integer to >describe (and similarly pjh:Paris is different from at least some Texan's >notions), but I don't consider that an adequate response. I think that we have to give a response like this, ultimately. Without this kind of start to the business, communication can never get off the ground. To be sure, it may turn out that when you and I talk about something for a while, it gradually dawns on one or both of us that either we disagree about things that it seems unlikely anyone could disagree about, or else we are using some words differently (for you and I, Lynn, "truth" was one of those words for a while). And at that point we may need to abandon, or put aside, the initial common ground and work instead on reconciling our terminology. I can imagine that one day, software agents on the SW may even engage in similar kinds of transactions. But we have to start somewhere, both human beings and machines. And the W3C has managed to put together what I think is a good workable set of conventions (including the idea of 'ownership' of urirefs, so yes, I *do* get to say what pjh:Paris means) which look like they at least can enable the world to get started on this business. If we insist that all communication is contextual and we can never know what anyone is saying, then we will never get started. If babies used this strategy they would never learn a language. One has to start by taking things on trust and maybe later come back and fix things when they go wrong. > There is NO definitive >objective interpretation of what, e.g., Paris mean Well, just as a philosophical aside, I disagree. But this really is a *philosophical* disagreement, you understand. >s, and although I'm confident >that you can constrain pjh:Paris in ways that would make it clear that >texas:Paris is a different thing, the same issues with subtler >shading will arise >over and over again. I have said before that the only way to truly understand >web ontologies is in terms of communities of interpretation -- >social context -- >and I will say it again now. OK, I can agree with that. Ultimately this may indeed be required for a full, true, real, understanding. But your next sentence seems to go in a completely different direction: > The trick is going to be the ability to say >something about being in the same social context and therefore using Paris to >mean France, not Texas. But this is exactly what I was talking about. If you say it, as part of the communication of your content, then its no longer part of the context. Yes, I agree, that is what we need: ways of SAYING which context (social or otherwise) we are speaking in. But notice, that once that is said, the whole assertion is NOT in that context; it has been decontextualized by including a reference to the context inside it. Seems to me that you have two choices here. One is insist that no matter how much we say about the intended context, that this in turn is contextual, and that any attempt to completely decontextualize an assertion is doomed by an infinite recursion. This is the radical position that leads to the nonstandard logics (and which I think is unlikely to produce a useful workable semantic web in any case , since these poor dumb software agents just arent smart enough for it to be worth getting this sophisticated about their social situation: at some point they are going to bottom out in something like elementary inference, and just stop worrying about contexts and semiotics. ) Or, you can say that we only need to incorporate enough of the context to make it clear what we mean; and in that case, you are arguing for exactly the same point that I was making. > (Said differently: If the trick involves its being TRUE >that Paris can only mean France, the trick ain't gonna work on the web.) Oh dear, I see that I have raised your ire by using the T word. I didnt mean to imply that asserting something made it Really True in any cosmic sense; only that asserting it amounts to a CLAIM that it is true; which is really no more than a transcription of the meaning of 'assert'. Im not saying that 'paris' can *only* mean the one in France. I am however saying that we can reasonably expect that any published assertion be clear which actual place it is referring to, and even that there could be a global presumption that when used without any qualification, names of capital cities refer to the capital city. I don't think it is reasonable to say that because all linguistic meaning is contextual and relative, that the basic machinery of the SW should be smart enough to figure out which city people are talking about from the 'context' of the RDF on their web pages (whatever the hell that could possibly mean.) On this view, for example, if Bubba had used 'paris' in his RDF, then he had just mis-spoke without realizin' it. It was up to him to say it right, not up to us to figure out his meanin'. And with regard to truth: in fact, yes, what he says had better be TRUE, or his packages aren't going to get to the right airport. At some point all this reasoning and meaning gets cashed out as action in the one, actual, real universe that we all inhabit. > >> >> (There is a possible reply to this, along the lines that nothing is >> every just plain true, and all assertions have to be understood as >> made relative to a surrounding context, which might itself be >> contextual or incomplete. I doubt if you want to go there, though, > >There is nowhere else to go. Well, then, Lynn, you are presumably opposed to the entirety of the current SW standards work (RDF, DAML, OIL and OWL) since NONE of this has made the slightest effort to go in that direction, and nobody has ever, to my knowledge, even suggested that we use anything like situation theory or context logics, even in a simplified form, as a web standard. That would be a very interesting kind of proposal, by the way, and I'd love to be spending my time on that kind of an idea rather than arguing over how to conform to XSD datatyping and whether qnames are acceptable in XML syntax. But I don't see a snowball's chance of any of these committees exploring in that direction. BTW, I disagree that there is *nowhere* else to go. We can get a lot done with a less-than-perfect solution. In fact, this is starting to happen. > >> What we (on these WGs) are trying to do is to design languages for >> use by agents on the semantic web, to be used to express content and >> transmit content across the web so that it can be used elsewhere. >> This presupposes, therefore, that the content expressed in these >> languages is being *published*, and once published, it might be used >> in ways, and by mechanisms, which are operating in contexts >> completely different from the context in which the content was >> originally stated; and moreover, those contexts are not known to the >> publisher of the content. > >In fact, we're not going to be able to know the context, but we WILL >be able to >(and need to) say: For the purposes of the following, assume that >you are in the >same context as the author of this information. That presupposes that the author has described the context as part of the information somewhere, however. If one has no way to recover the context of the publication, this 'assumption' is impossible. >Your assumption may be wrong -- >people coopt terminology and misuse it all the time The issues of misuse and cooption are orthogonal to the point we were discussing. I agree that are a host of issues connected with one person making assertions using terminology define by someone else. But leave those aside for now. >-- but your reasoning will >need to be based on this presumed sharing of context. That is like saying that an SW reasoner has to be telepathic. HOW is an agent supposed to base its reasoning on an unknown context? > For the most part, that >presumed sharing of context will have to be indicated by human >beings or webs of >trust. I have no idea what this is supposed to mean. What is a 'web of trust' ? Can one use HTTP to send trust around? > >> Given this, it follows that publishing any >> assertion which is meant to be understood in a particular context, >> and only means what it is supposed to mean when that context is >> shared by its reader and its publisher, is at best misleading and at >> worst irresponsible. We cannot presuppose any - or at least, more >> than a tiny amount of - shared context, so our web languages must >> make all - or at least, as many as possible - of their contextual >> presuppositions explicit. They need to be decontextualized. > >I agree with almost all of your reasoning about what is or isn't >possible on the >web but draw somewhat different conclusions. I agree that maximum possible >decontextualization is a Very Good Thing, so I guess I don't >disagree with most >of this text (including large portions I've omitted from this >response). But I >fundamentally doubt that decontextualization is POSSIBLE, in general. Maybe not, in some absolute sense. But I think that this very strong sense isn't a useful place to start; and for sure, it has nothing to do with the design of RDFS/DAML/OWL/OIL. >That is, >even with maximal decontextualization, we're going to need the web >of trust/it's >OK to just assume you're in the same context as this ontology. Seems to me that this kind of talk only starts to be useful when we come to consider how to deal with what happens when the 'web of trust' breaks down. I agree this is very important area and I wish more people were worrying about it; but in the meantime, do you really want to be arguing that it is OK for someone to publish an SW assertion that the universe has only 7 things in it, just because all assertions are ultimately contextual? That seems to me like arguing that since no human being is morally perfect, that it should be OK to commit murder. > >> >> >> PS. that term, 'universe of discourse', isn't just an empty label. >> The pioneers of formal logic knew very well that logics formalize an >> aspect of language use, and that all linguistic meaning depends on a >> kind of contract involved in all human language use, which is a >> shared understanding of what the communication is about, the 'common >> ground' of a discourse (or of a mathematical proof, for example). > > Apply the idea to human telephone conversation. If I feel in my >> pocket and say to you, only seven things exist, then in order for us >> to continue to share our assumptions - to have a common ground - you >> have to accept this rather questionable claim. Which is unlikely if >> you are a sane adult, and irresponsible of me if you are sufficiently >> gullible. But if I say, there are seven things in my pocket, then the >> chances of our being able to go on communicating successfully are >> greatly increased. > >Fine, but logic presumes there is *A* universe of discourse No. Each *use* of logic presumes that there is a universe of discourse. But of course the same logic, and even the same logical vocabulary, can be used to describe a variety of such domains. My point was that (until we have ways of negotiating and brokering meanings, which has to lie in the future) web *publication* amounts to a particular kind of use: it broadcasts an assumption into a kind of global conversation. Which requires everyone to use a little care if this global conversation is not to break down in confusion. Imposing a constraint on the planet-wide universe of discourse might have unintended consequences which are potentially dangerous. Web publication should be considered hostage to changes of context, and therefore should be as decontextualised as possible; and in particular, that glaring examples of contextual dependence which impose a ludicrously strong condition on the universe of discourse, such as a claim that is true only inside C code, or true only within a database context, should not be published on the WWSW. (BTW, I know you think of logic as a demon that needs to be exorcised; but really, you should be careful. If all logic is exorcised from the semantic web you may not like what you are left with.) >and, unfortunately, >on the web there are many localized universes that partially overlap >and slip and >slide against one another. That is, you can't decontextualize >everything so that >there can be a single universe behind it all. I agree with your point (though not your way of expressing it, which I vehemently reject: there IS a single universe behind everything. Another purely philosophical debate.) But I also think that the SW is only going to work, or anyway only going to get started, if it is built on the assumption of this fiction, even if it is fictional. And I think that human society and human communication is only possible because we all accept a similar fiction. In some deep sense it is probably impossible to fully communicate a thought using words; there will always be aspects of meaning inside your head which I don't get right in my head. We are all a local context of one. All the same, we do get by most of the time very well, particularly when it comes to what one might call the pragmatic everyday business of life. We need to find a similar compromise for the SW. > And if you think you can, you're >missing the idea that mutually inconsistent pockets of reasoning >happen *all* the >time in the world (not to mention within a single human brain, naive >theories of >belief notwithstanding). I know, I know. I read about stuff like that occasionally. (There is also a large body of empirical evidence for a number of internal strategies in our heads whose sole purpose is to remove those pockets, by the way, even to the point in some cases of producing bizarre beliefs in the process; postmodernist speculations on the nature of meaning notwithstanding.) But I'm also enough of an engineer to see that at some point we have to just try building this thing, perhaps with false assumptions and crude simplifications built into it. Once it exists, other people can make it better. All I was arguing with Ian and Ben about was that when we do this, we at least make an *attempt* to think about the larger issues. Pat -- --------------------------------------------------------------------- IHMC (850)434 8903 home 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32501 (850)291 0667 cell phayes@ai.uwf.edu http://www.coginst.uwf.edu/~phayes
Received on Sunday, 29 September 2002 18:08:18 UTC