It's WYSIWYG if you think of the document in an abstract sense, separated from all style (or in your own style -- knowing that others will see it in their style).
Even a bulleted list needs a style. Certainly paragraphs do. Note how many options there are in CSS for controlling the style of each one.
In the original web, the idea was more that such style would be determined by the reader (via the browser), than by the writer/content-creator.
Note that in a Wiki the author has no control over the style -- either there is a site-wide style, or the reader gets to pick a style for all the pages. Same idea for the original web.
I'm not sure what point you're trying to make about the relationship between RDF and Model Theory. Does it really matter to what extent RDF is "driven by model theory"?
I think the RDF Semantics specification was written in a manner illuminated by MT, but the basic design and utility of RDF are not particularly related. They come more from frame systems, semantic nets, and the like; it's a small fragment of first-order logic.
If you'd actually like to engage in productive discussion about RDF or the relevance of the W3C, you might try some of the W3C mailing lists.
(Disclosure: I'm part of the "RDF staff", although we didn't write the specs; mostly the staff just offers support for the members of the Working Groups who do the real work. Of course we do accept some responsibility for problems in the spec, so if you'd care to explain better, I'll listen.)
The phrase "well-defined meaning" is supposed to convey to experts that these languages are formal languages (with formal syntax defined by grammars and formal semantics described using model theory), while staying within the general vocabulary appropriate for a press release or Scientific American.
I'm not thrilled with the phrase, myself.
My offering: The Semantic Web is the part of the Web where information is conveyed not in natural language or proprietary and legacy formats, but in languages designed so web clients can effectively gather, combine, and perform application-specific processing on information from servers around the web.
This connects with his mistaken point that the Semantic Web is based on some single universal ontology. This is of course the opposite of what RDF is about -- it's about allowing lots of ontologies to be used side by side.
So we don't model the real world perfectly, we model it well enough for some set of applications in some ontology. Every database designer, nearly every programmer does this all the time. We model it well enough and then the computers... do what computers do.
RDF is nothing new here. What's new is establishing a fairly wide and precise consensus around a language for communicating data about arbitrary things.
Okay, I thought those were relatively pleasant reads, which can be a plus. (And I wanted to say something fast, before slashdot buried any response I might make...)
My actual response at the time is brief and chatty. The response from Dan Brickley is also short and sweet. Neither of us felt it was worth the time to reply point-by-point.
The "misquoting" is to suggest that my "how you buy a book on the Semantic Web" sketch should possibly cause Jeff Bezos to lose sleep. I was trying to explain an experimental protocol in a way I hoped my grandmother could understand (seriously!) and Shirky thinks I'm sketching out Amazon's doom? I don't expect the Semantic Web to doom anyone but folks who want to keep data exchange laborious.
RDF/XML uses XML namespaces as a somewhat convenient way to write URIs (which are normally quite long).
RDF (in the abstract) doesn't use namespaces, it just uses URIs (aka URLs). (The concept of namespaces is still there in effect, as a collection of related names, in an ontology -- but that's quite different from the formalism of XML Namespaces.)
Shirky's peice on the Semantic Web is far below his normal quality. It's poorly researched and poorly considered. (Speaking as someone misquoted in the article...)
While I'm all for better markup, there's quite a jump from proper use of "semantic markup" in HTML to RDF. RDF is quite another language.
Semantic Web, Ontology Mapping, Trust
on
Sentient Data Access
·
· Score: 5, Insightful
XML only solves the first problems of data merging, like making the formats easy to parse correctly and using the right international character set.
RDF/XML solves a bit more of the problem, making the structure of the information clear, in terms of assertional statements. An RDF/XML file is a knowledge base, full of statements saying this has some particular relationship to that. It lets the machines get at more of the information in a uniforn, universal way.
But still, the problem of ontology/schema/vocabulary mapping remains: if one system is talking about patients and another is talking about clients, they might or might not really be talking about the same thing. A single person maybe never counts as two patients but sometimes counts as two clients, etc. At least with the data in RDF, most of this mapping can be done in software once a person figures it out and expresses it in a suitable logic language.
The emerging design of the Semantic Web hopes to make that reasonable, but also to support convergence on common vocabularies by having everything on the web -- if it's trivial to see what vocabularies are already being used, people will mostly only make new ones when the old ones really are different.
Other hard problems remain, of course, like figuring out which data sources to trust. Fun fun.
Actually, I think all parents are sensitive about being told how to raise their children. I imagine it's because it's so hard, and impossible to do perfectly.... Seems like a recipe for sensitivity to me.
Thanks for the mention (I'm co-author of the tag: URI draft), but tag: URIs are languishing in large part because it's not that clear they are useful. They have some of the same problems as info:, namely they don't give you any handy functionality. They just kind of sit there. They're not not nearly as cool as http, for most uses.
That said, I still kind of want them approved to help focus people who DO want this kind of thing. But since it's not very important to me, I have a hard time putting time into jumping IETF hoops.
Re:RDF is quite pratical
on
Practical RDF
·
· Score: 1
Um, while I'm a big supporter of RDF (it pays my salary), I don't even know of one website which explains well how to use RDF, in general. Yes, for RSS, probably, but RDF in general?
Isn't CDDB going private with user-submitted
data the same thing as happened with IMDB going
private with data that
users submitted (and they used to allow you
to download)...?
I see the point that they too put a lot of work into the database, but I think it's abhorent (if not actually illegal) to neither warn the users who submit the data (impossible in these cases) nor make the raw data available for a fee no greater than their actual costs and subject to no license.
But I don't know what anyone can do about it, except refuse to submit data to sites that don't make their intentions clear and legally binding up-front.
With rsync you have to remember where you last made your changes. Use something much smarter, but using the same diff/compression protocol, like
unison
And that's a problem we could also address with an open-database version of slashdot.:-)
Yes, it (building good group-moderated systems) is a Hard Problem. But right now we're not allowed to even try to solve it without starting over from scratch again.
I want everyone to have open access to Slashdot's database of comments, so we can write a better (eg not web-based) UI. Of course if we had this access, Andover would lose some of its hold over its revenue stream. Still, it could license the data with a provision that viewing software must include Andover-specified ads and/or pay a fee. (That's hard to enforce, but nothing stops you from using ad-striping software now.)
Of course this is a special case of my general desire for API/open-protocol access to all databases. Why should I have to use Amazon's web-based interface to buy a book from them?
Right now the problem is both human and technical. XML will address some of the technical issues. (I'm working on some others.) The political and commercial issues will be tricky.
I know full-well that "root" can be pronounced at least two different ways, but since it didn't matter to my arguments I didn't try to distinquish between them. Wired said: "But [Gore] mispronounced 'routers' as root-ers" and since I can't tell which way they meant, I just stuck with their form. I should have quoted it, attributed it, and marked it with "sic," I supposed.
Re:In defence of Gore, Prescriptivism(standards)
on
Tales From The Bazaar
·
· Score: 1
It sounds like we're in violent agreement. I was attacking McCullagh/Wired for their attack on Gore's saying "rooter". I think there are at least two completely legitimate ways to prounounce "router" and Gore chose one of them. That Wired didn't know both pronounciations were okay appalled me.
Personally, I'm deeply in the descriptive linguistics camp, although becoming less so as I try to convince my three-year-old to use prounciations that other people can understand. Still, his slightly-off pronounciations of some words is very cute, like compoooooter (very long 2nd syllable, missing "y" sound) and "Electra City" (for electricity).
BTW, I think coming up with unambiguous and obvious ways to convey pronounciation in ascii is a wonderful challenge. ASCII IPA is nice, but fails the obviousness test.
Slashdot Mirror
Yes, what Tim means by WYSIWYG (I'm pretty sure) is immersive editing, a direct manipulation interface.
It's WYSIWYG if you think of the document in an abstract sense, separated from all style (or in your own style -- knowing that others will see it in their style).
Even a bulleted list needs a style. Certainly paragraphs do. Note how many options there are in CSS for controlling the style of each one.
In the original web, the idea was more that such style would be determined by the reader (via the browser), than by the writer/content-creator.
Note that in a Wiki the author has no control over the style -- either there is a site-wide style, or the reader gets to pick a style for all the pages. Same idea for the original web.
I'm not sure what point you're trying to make about the relationship between RDF and Model Theory. Does it really matter to what extent RDF is "driven by model theory"?
I think the RDF Semantics specification was written in a manner illuminated by MT, but the basic design and utility of RDF are not particularly related. They come more from frame systems, semantic nets, and the like; it's a small fragment of first-order logic.
If you'd actually like to engage in productive discussion about RDF or the relevance of the W3C, you might try some of the W3C mailing lists.
(Disclosure: I'm part of the "RDF staff", although we didn't write the specs; mostly the staff just offers support for the members of the Working Groups who do the real work. Of course we do accept some responsibility for problems in the spec, so if you'd care to explain better, I'll listen.)
Why do you need more video streams than there are people in the house?
Where in the specs do you find any suggestion that a reasoner should make the closed-world assumption? I don't think it's there.
The phrase "well-defined meaning" is supposed to convey to experts that these languages are formal languages (with formal syntax defined by grammars and formal semantics described using model theory), while staying within the general vocabulary appropriate for a press release or Scientific American.
I'm not thrilled with the phrase, myself.
My offering: The Semantic Web is the part of the Web where information is conveyed not in natural language or proprietary and legacy formats, but in languages designed so web clients can effectively gather, combine, and perform application-specific processing on information from servers around the web.
Still a mediocre sentences, sorry.
This connects with his mistaken point that the Semantic Web is based on some single universal ontology. This is of course the opposite of what RDF is about -- it's about allowing lots of ontologies to be used side by side.
So we don't model the real world perfectly, we model it well enough for some set of applications in some ontology. Every database designer, nearly every programmer does this all the time. We model it well enough and then the computers... do what computers do.
RDF is nothing new here. What's new is establishing a fairly wide and precise consensus around a language for communicating data about arbitrary things.
The OWL sense of "ontology" is the second sense, if you read "theory" in the formal (computer science/mathematical logic) sense.
That is, an OWL ontology tells readers (especially computers) what kinds of things exist and what kinds of relationships they can have to each other.
Some of the OWL specs are actually pretty readable. Try starting with the OWL Overview. (Others, like OWL Semantics, are... more challenging.)
Okay, I thought those were relatively pleasant reads, which can be a plus. (And I wanted to say something fast, before slashdot buried any response I might make...)
My actual response at the time is brief and chatty. The response from Dan Brickley is also short and sweet. Neither of us felt it was worth the time to reply point-by-point.
The "misquoting" is to suggest that my "how you buy a book on the Semantic Web" sketch should possibly cause Jeff Bezos to lose sleep. I was trying to explain an experimental protocol in a way I hoped my grandmother could understand (seriously!) and Shirky thinks I'm sketching out Amazon's doom? I don't expect the Semantic Web to doom anyone but folks who want to keep data exchange laborious.
Microsoft didn't have people active in either Working Group. They didn't fund this any more than any other W3C (Full) Member.
Why store them separately?
Anyway -- sounds like an excellent project. I'm not aware of anyone doing it quite yet.....
RDF/XML uses XML namespaces as a somewhat convenient way to write URIs (which are normally quite long).
RDF (in the abstract) doesn't use namespaces, it just uses URIs (aka URLs). (The concept of namespaces is still there in effect, as a collection of related names, in an ontology -- but that's quite different from the formalism of XML Namespaces.)
Shirky's peice on the Semantic Web is far below his normal quality. It's poorly researched and poorly considered. (Speaking as someone misquoted in the article...)
For good responses see Peter Van Dijck or Paul Ford.
While I'm all for better markup, there's quite a jump from proper use of "semantic markup" in HTML to RDF. RDF is quite another language.
XML only solves the first problems of data merging, like making the formats easy to parse correctly and using the right international character set.
RDF/XML solves a bit more of the problem, making the structure of the information clear, in terms of assertional statements. An RDF/XML file is a knowledge base, full of statements saying this has some particular relationship to that. It lets the machines get at more of the information in a uniforn, universal way.
But still, the problem of ontology/schema/vocabulary mapping remains: if one system is talking about patients and another is talking about clients, they might or might not really be talking about the same thing. A single person maybe never counts as two patients but sometimes counts as two clients, etc. At least with the data in RDF, most of this mapping can be done in software once a person figures it out and expresses it in a suitable logic language.
The emerging design of the Semantic Web hopes to make that reasonable, but also to support convergence on common vocabularies by having everything on the web -- if it's trivial to see what vocabularies are already being used, people will mostly only make new ones when the old ones really are different.
Other hard problems remain, of course, like figuring out which data sources to trust. Fun fun.
Actually, I think all parents are
sensitive about being told how to raise
their children. I imagine it's because
it's so hard, and impossible to do perfectly....
Seems like a recipe for sensitivity to me.
Thanks for the mention (I'm co-author of the tag: URI draft), but tag: URIs are languishing in large part because it's not that clear they are useful. They have some of the same problems
as info:, namely they don't give you any handy functionality. They just kind of sit there. They're not not nearly as cool as http, for most uses.
That said, I still kind of want them approved to help focus people who DO want this kind of thing. But since it's not very important to me, I have a hard time putting time into jumping IETF hoops.
Um, while I'm a big supporter of RDF (it pays my salary), I don't even know of one website which explains well how to use RDF, in general. Yes, for RSS, probably, but RDF in general?
Show me the websites.....
I see the point that they too put a lot of work into the database, but I think it's abhorent (if not actually illegal) to neither warn the users who submit the data (impossible in these cases) nor make the raw data available for a fee no greater than their actual costs and subject to no license.
But I don't know what anyone can do about it, except refuse to submit data to sites that don't make their intentions clear and legally binding up-front.
With rsync you have to remember where you last made your changes. Use something much smarter, but using the same diff/compression protocol, like unison
Dr. Norman Matloff's web site, giving a great deal of information and opinion on the matter, is here, at http://heather.cs.ucdavis.edu/itaa.rea l.html.
Yes, it (building good group-moderated systems) is a Hard Problem. But right now we're not allowed to even try to solve it without starting over from scratch again.
Of course this is a special case of my general desire for API/open-protocol access to all databases. Why should I have to use Amazon's web-based interface to buy a book from them?
Right now the problem is both human and technical. XML will address some of the technical issues. (I'm working on some others.) The political and commercial issues will be tricky.
I know full-well that "root" can be pronounced at least two different ways, but since it didn't matter to my arguments I didn't try to distinquish between them. Wired said: "But [Gore] mispronounced 'routers' as root-ers" and since I can't tell which way they meant, I just stuck with their form. I should have quoted it, attributed it, and marked it with "sic," I supposed.
Personally, I'm deeply in the descriptive linguistics camp, although becoming less so as I try to convince my three-year-old to use prounciations that other people can understand. Still, his slightly-off pronounciations of some words is very cute, like compoooooter (very long 2nd syllable, missing "y" sound) and "Electra City" (for electricity).
BTW, I think coming up with unambiguous and obvious ways to convey pronounciation in ascii is a wonderful challenge. ASCII IPA is nice, but fails the obviousness test.