W3C home > Mailing lists > Public > public-sw-meaning@w3.org > May 2004

Re: Comment on "Named Graphs, Provenance and Trust"

From: Pat Hayes <phayes@ihmc.us>
Date: Tue, 11 May 2004 15:04:58 -0500
Message-Id: <p06001f54bcc6d24f10d9@[10.0.100.76]>
To: Bijan Parsia <bparsia@isr.umd.edu>
Cc: public-sw-meaning@w3.org, "John Black" <JohnBlack@deltek.com>

>On May 11, 2004, at 1:22 PM, Pat Hayes wrote:
>
>>>On May 6, 2004, at 3:11 PM, Pat Hayes wrote:
>>>[snip]
>>>>>This is what I previously referred to as a 'stipulative definition' or
>>>>>'stipulative ontology'.  What I am trying to get at intuitively is the
>>>>>ability to say, "When this term is used in this context it SHALL be
>>>>>interpreted to mean that".  And as a crude example of its use, imagine
>>>>>I had need to import two ontologies, one a universal business language
>>>>>ontology of commercial transactions, and the other, Bijan and Peter's
>>>>>alternative ontology of transactions, both of which contained some terms
>>>>>I needed but which had an incompatible "invoice" term that caused an
>>>>>ambiguity (inconsistency).  Now I want to be able to eliminate the
>>>>>ambiguity by stating that within this named graph, when the term
>>>>>"invoice" is used, it SHALL be interpreted according to the UBL ontology
>>>>>(or vice versa).
>>>
>>>(The above was from John Black.) Interestingly, John, this is 
>>>exactly the kind of scenario we were interesting in making 
>>>possible, even though the current tools (owl:imports, cut and 
>>>paste, etc.) are a bit crude. The current canonical way to achieve 
>>>this is to simply only include in your document the desired 
>>>axioms. On some views (e.g., some variant's of Tim's view), it was 
>>>much more difficult to exclude the URI owner's axioms.
>>
>>I don't think that interpretation was ever Tim's intended meaning 
>>(I.e. the interpretation in which importing the URI owner's axioms 
>>was kind of obligatory).
>
>I do. At least, I have elicited such view from Tim (and distinct 
>views...his line is much softer on this for quite a while), and from 
>other people.

Ive never heard Tim say this (and in some cases he was arguing with 
you at the time.) What he did say was that the correct meaning of the 
URIs was to be specified by the owner of the URI (and accessible 
somehow from the URI itself). That is not the same as talking about 
importing. And I also think that the intention of stating even this 
much as part of a spec was not to IMPOSE meaning so much as to 
provide a system-endorsed way to DETERMINE INTENDED meanings in the 
case where a reader finds a URI used without further explanation. 
And even there, after taking part in some of the TAG debates 
concerning the architecture document, I think that what might be 
called 'rules of good practice' can be part of a spec and still 
intended to be understood in much less authoritarian a way than could 
possibly justify all this use of rhetoric like tyranny and oppression.

>>You only get to that from what Tim was (I think) talking about by 
>>imposing a very limited view of what 'meaning' means.
>>
>>>However, as pat point out, you can't stop me (currently, or 
>>>really, ever) from adding yet more axioms to your set.
>>
>>Wait a minute. Look, a SET is a mathematical abstraction. It doesnt 
>>make sense to speak of DOING something to a set.
>
>Sigh. I meant Python Sets, which are mutable :)

Oh, well then in that case you can't add to my set. You can copy it 
and add to the copy, which is yours.

>>What you can add axioms to is a document or an ontology or some 
>>concrete thing on the Web somewhere. So lets talk about those. I 
>>CAN stop you adding axioms to MY document, and of course you can 
>>write documents of your own out the wazoo, and can incorporate my 
>>axioms into them (by reference or by cutting and pasting) and I 
>>can't do anything about that.  All that is obvious.
>
>Strange how it hasn't always been to some people in some conversations.

I think it always has, in fact. I have never heard anyone deny any of 
this. Like I said, I think you were misunderstanding the intentions 
of the proposals.

>>>E.g., if I import *your* document, I'm almost surely going to add 
>>>assertions, and I'm free to add axioms, even concerning your 
>>>classes.
>>
>>Right, of course (does this ever need to be said again??)
>
>Yes. Or, at least it's possible that it will have to be said again.
>
>I was just starting from the obvious case. Trying to start from a 
>point I felt confident we all agree on.
>
>>but YOU are saying those axioms, (in YOUR document), right?
>
>Yes.
>
>>John isn't asserting them. So take a hypothetical third party, call 
>>him Joe, who can read all this stuff. If Joe reads what Bijan 
>>publishes he will get a potentially different idea than the one he 
>>would get if he only reads what John publishes. Of course: so what? 
>>Does that change the meaning of what John published? No. Does it 
>>change the meaning of John's terms as used by Bijan? Seems to me 
>>it's up to Joe to decide that. If Joe believes Bijan as well as 
>>John, then the meaning of the terms is determined by (Bijan's 
>>ontology + John's ontology) as far as Joe is concerned. What if 
>>John protests that his terminology is being misused by Bijan. Well, 
>>he may be right.
>
>Sure. Though, I don't want to be told that my misuse is a matter of 
>Web/RDF/OWL architecture or standards.

I think its not unreasonable for such standards to provide John with 
a way to establish that your use is misuse, by his standards. In this 
case the dispute about meaning is between you and John, and all the 
standards do is to allow you two to express your intentions more 
precisely.

>I want to be able to do the analogue of deep linking. I want to use 
>your URIs and make a slew of assertions involving them.

Nobody can stop you doing that, for sure.

>If you've, from my perspective, published crap about the URI at some 
>relevant web location, I don't want to have any technical pressure 
>at all for including it.

There is none.  Nobody is talking about inclusion. But there might be 
a way for someone else to make their difference of opinion with you 
evident to anyone who uses the URI to access the owner's ontology. 
That is, ownership of a URI should be able to count for something. 
The world will have to decide this for itself, but I'd predict that 
normal social pressure would follow this general rule, if we provide 
for some way to assert things clearly and unequivocally.

>I don't want to have to keep delving into nasty preference files, etc. etc.

I have no idea what you are talking about. What proposal has this implication?

>I think this inhibits reuse, personally.

I think there is a likely use case where a publisher publishes what 
is little more than a list of URIs, with the intention to do little 
more than encourage re-use and let their final meaning, if this term 
even has a meaning itself, be determined by their wider use. For 
this, I think we all agree that the conventions that are announced 
about importance or otherwise of author's intentions will be 
irrelevant. So this can always be done. But to assume that this must 
be the norm is like assuming that the entire Web consists of blogs.

>
>>  Still, Joe can reply: OK, but I'm not only hearing you, I'm also 
>>hearing Bijan. Suppose John admits that there's no way that he, 
>>John, can stop Joe listening to Bijan, but still wants to restrict 
>>the meanings of his terms *as intended by him* to be only the 
>>meaning that can be inferred from his, John's, publication.
>
>In his document.

Right.

>Neither Peter nor I have ever argued against anything like this. 
>We've argued for it. But we want it to be the case for all document 
>authors. I say what I say.

That is too simple, for several reasons. First, many URI owners might 
deliberately not constrain their intended meaning very strictly, in 
order to facilitate re-use (I would put the DC stuff in this 
category, for example.) Second, there will be cases where a document 
is not intended to be a positive assertion, and/or where the owner 
might want to explicitly not endorse the content. Third, we may need 
to negotiate things like claims and counterclaims. We have to allow 
for more nuanced expressions of intent of a publisher to the content 
published.

>If you want to combine what I say with what the "word (uri) owner" 
>says, that's *your* lookout.

Agreed.

>
>>That is about the best that he can do, seems to me, and that is 
>>what our proposal allows him to do. It would enable Joe, if he were 
>>so inclined, to discover that Bijan was using John's terms in ways 
>>that violate John's intentions, and so (if he, Joe, were so 
>>inclined) to follow John's intentions more accurately: it allows 
>>Joe to resist the infection by Bijan of alien meaning into Joe's 
>>understanding of John's vocabulary: again, all assuming that Joe is 
>>interested more in what John says than in what Bijan says.
>>
>>However, all that said, it seems to me that what might be called 
>>the normal case is one where everyone is kind of in broad agreement 
>>about intended meanings, and everyone acknowledges that in a case 
>>of a clash of meaning or intention, that the 'owner' or originator 
>>of a URI has more social authority over the intended meaning than 
>>any other single authority.
>
>I think this is insupportable. I really do. Especially in the 
>absence of the normal overrides for that authority.

Its not a question of support. Its a question of what will in fact 
turn out to be the normal case, in practice. I was making a 
prediction, not proposing a specification.

>
>Not in every case, but in many cases, especially casual reuse. Dan's 
>example is compelling.
>
>>There will be exceptions, eg if an entire community 'misuses' a 
>>term in some way consistently, but such phenomena are familiar 
>>already and we know, broadly , how to handle them.
>
>Right. I don't want to preclude them with hamfisted legislating. I'm 
>the quietist on all this, after all.

Interesting: I don't get that impression. Perhaps we are not 
communicating adequately.

>  Though I don't object to the kind of moves you make in your paper, 
>in principle.
>
>>>Even if I just add facts ("instances"), I can affect the 
>>>*inferred* axioms (e.g., the classification). So the question 
>>>becomes, how far reaching should your stipulations reach?
>>>
>>>I *like* being able to say, of my documents, that the axioms are 
>>>*these* ones I put in some effort to put in, and not *those* which 
>>>I put in some effort to exclude.
>>>
>>>>But the performative story only gives you a kind of act, not a 
>>>>stipulation about other's acts or interpretations. So let me try 
>>>>to rephrase this a little: suppose we say that the intuitive 
>>>>force of such a stipulative definition is to declare that *your 
>>>>intention* is that the terms shall have the meaning you give to 
>>>>them, and that any usage (such as Bijan and Peter's, in the 
>>>>example) which clashes or is incompatible with your meanings is 
>>>>not intended by you. This of course does not *prevent* P & B from 
>>>>misusing your terminology (how could it?) , or force anyone else 
>>>>to believe you rather than them (so all this foolish
>>>
>>>Foolish stupid? Foolish tactically unwise? Foolish cute?
>>
>>Foolish stupid, since you put it that way. I think y'all are 
>>misunderstanding the point.
>
>Likewise, evidently :)
>
>>Take the http GET protocol. This sets out detailed instructions 
>>about exactly what must be done by browsers, network software and 
>>servers when you click on a hyperlink in a Web page. This stuff is 
>>intensely proscriptive: so does the HTTP spec oppressively limit 
>>the freedom of software to act as it chooses? Is this tyranny in 
>>action? Well, if you like, you could say that, but this would be a 
>>foolish conclusion.
>
>No, but there are differences that make a difference.

Maybe it would help if you articulated them. To me, what you say on 
this theme reads pretty much like a direct transcription of this kind 
of rhetoric from the HTTP protocol to the OWL protocol, and it makes 
no more sense in either one. We are in both cases talking about 
software, right? What is this talk of tyranny and freedom supposed to 
mean when applied to software?

>Clearly both Peter and I are find

find/fine ?

>with normative specification and are totally inclined to be rather 
>insistent about what you say an OWL document means. I find it to be 
>a recurring irritation, going all the way back to the Boston tech 
>plenary, that I am the object of such imputation. There *are* people 
>who argue against standardized vocabulary in general, quite 
>strongly. I'm not one of them, clearly.

Im afraid I do not follow what you are saying here, what it is that 
you think I am imputing to you. Did I say anything about standardized 
vocabularies?

>>  In fact, it is false: I *could* write a browser which claimed its 
>>right to treat hyperlinks differently, cast off the oppressive 
>>chains of the HTTP protocol and act in its own proud 
>>individualistic way.
>
>Pfft.
>	or
>:)

Both, I hope. But seriously, it seems to me to be no more ludicrous 
to use this kind of rhetoric here than in the Sweb case where you and 
Peter use it.

>>It would just be a useless browser, of course, rather in the way 
>>that human being who insisted on not believing anything told to it 
>>would be a useless human being (we all know some of these, right?) 
>>and a SWeb agent which refused to draw conclusions from any 
>>ontologies except those directly indicated would be useless. The 
>>point is not that any of these things are OBLIGED by some external 
>>tyrannical compulsion to conform to any social convention, so much 
>>as the observation that if they don't conform then they aren't 
>>going to be able to work very effectively, particularly if everyone 
>>else is conforming: which is the entire point of having the 
>>conventions in the first place.
>
>We're worried about the nature of the conventions proposed. Some irk 
>us and we think will irk others. They will be oppressive at least in 
>the sense that some people (e.g., many Lispers) find programming in 
>C++ oppressive.

?? I don't follow you. Are you saying that some set of conventions 
imposes a particular programming style?

>  There is room, however, for worse on the semantic web.
>
>>  If you approach any social system of communication with a 
>>basically paranoid assumption that everyone is out to screw you,
>
>It's not unreasonable to be concerned with various senses of 
>security in a social system of communication, in *any* one.

True; but it is, I seriously suggest, a mistake to assume that issues 
of security and protection from misuse are the PRIMARY concern of a 
set of conventions intended to facilitate communication. First lets 
figure out how to transfer and communicate meanings, then worry about 
security later. If you can't even talk, worrying about lying is 
premature.

>There are special issues in a system involving significant automated 
>processing in the midst of that communication.
>
>>  and with the prime goal of avoiding infection by other's ideas and meanings,
>
>We're trying to avoid a situation where we feel compelled to act that way.

Frankly, I would be inclined to suggest that if you really do feel so 
compelled, then  by all means don't use the semantic web; just as if 
you are more worried about catching a cold than anything else, don't 
go out and talk to people. The rest of us will probably get along 
fine most of the time.

>>then you are likely to not be able to communicate as effectively as 
>>an agent which approaches communication with the intention of 
>>communicating. You and Peter are adopting a stance of what might be 
>>called semantic paranoia as a basic operating assumption of the 
>>SWeb, which seems to me to get the entire thing backwards.
>
>We argue for choice.

Choice for the reader, yes: but that does not need to be argued, 
seems to me, since readers have freedom, by the very nature of Web 
architecture. But you reject choice for the writer and owner to 
express their clear intentions, which seems to me a more corrosive 
kind of rejection. If all SWeb ontologies just kind of appeared, like 
fruit on a tree, and all we had to worry about was using them 
appropriately, then your position would make sense: but if we think 
of the Sweb as embodying a kind of communication, then the issue of 
how a reader can discover the intentions of the writer 
(owner/publisher/whatever) becomes more central.

>I don't believe we do argue for semantic paranoia. If you want to 
>import everything under the sun, go ahead.

Thanks, I will. But then I guess I fail to understand what your point 
is in this debate.

>It seems to be a sane and familiar way to operate. Or if you are 
>writing processing software, aggregated a bunch of claims into a kb 
>and work on that.
>
>>>>analogizing with tyranny, oppression and freedom is beside the point)
>>>[snip]
>>>
>>>I dispute that it's (necessarily) a misuse, so perhaps the 
>>>analogy, about which I was having a few qualms, is apt.
>>>
>>>The point of this discussion is how much de jure authority we are 
>>>giving to URI owners.
>>
>>None whatsoever, in the sense that you seem to be understanding 
>>'authority'. No amount of web publication in any medium obliges any 
>>reader to endorse, utilize or agree with any of it if they choose 
>>not to: and this goes for software as much as it does for human 
>>agents.
>
>If OWl, for example, required some form of ontological closure 
>beyond the explicit on of owl:imports,

Can you document ANY claim by ANYONE that this should be *required* 
by any form of compliance? To say that meaning is partly defined by 
something is not to say that any reader is obliged to assent to the 
meaning or to do anything about it, still less that any piece of 
software is obliged to do anything about it. The OWL specs don't 
require that software perform any inferences at all: a conforming 
engine can just read OWL files and ignore them.

>then OWL compliant software will process my documents a certain way. 
>I can, of course, avoid such OWL compliant software and perhaps 
>require people not to process my document with OWL compliant 
>software (at least, if they want me to accept the results). This 
>seems worse than doing things the other way around, frankly.
>
>>  This is just obvious. (So if someone seems to be saying something 
>>that contradicts this, maybe you should contemplate the possibility 
>>that you may have misunderstood what they were really saying.)
>[snip]
>
>Well, I was tempted to make a plea for charity several times 
>earlier, but resisted. But frankly, this is beyond my willingness to 
>bear at the moment. I'm totally surprised you would think that I 
>need this advice.

Well, sorry if it came across too strongly. It was based on what I 
still think is a basic misunderstanding you have of what I take to be 
Tim's position.

Pat


>Cheers,
>Bijan Parsia.


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes
Received on Tuesday, 11 May 2004 16:21:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:42:16 GMT