research topics

[Algernon]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

I will be studying for my PhD starting this fall. The program stresses
research. I have been trying to think of some areas and topics but am coming
up blank. Does anybody have any ideas for some research topics; AI in
general and Prolog in particular? And what are some periodicals,
institutions that publish research papers.

Thanks

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

You might look at the papers of the annual AAAI conference on the web.
I find these somewhat dry, because most of them seem to be 3rd-order topics.
But these guys will be your peer-reviewers for funding and publication.
I'd recommend traveling to San Jose, CA in four weeks to catch this year's.

Your graduate advisor should be the main source of topics.
S/he should know what is an important, do-able project. S/he should
protect from doing something thats already done or from working on
a project that will take ten years. Even if the topic is not perfect,
your advisor is the main gateway to passing the preliminary tests
and final defense, and the source of funding.

 

[Randy]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Algernon wrote:
> I will be studying for my PhD starting this fall. The program stresses
> research. I have been trying to think of some areas and topics but am coming
> up blank. Does anybody have any ideas for some research topics; AI in
> general and Prolog in particular? And what are some periodicals,
> institutions that publish research papers.

Your PhD program will have you taking classes for over a year before you start
doing any research. In fact, several years of coursework will probably precede
research.

Also AFAIK, nobody is doing research into prolog. It's a language based
entirely on simple horn clauses. As such, I can't imagine why it would merit
further development, especially academically. In fact, I doubt there's a lot of
research being done in practical logic-based AI either, except perhaps as part
of planning systems. In all likelihood, if you want to work in the logic side
of AI, you'll be doing mathematical proofs, not implementing systems or
extending programming languages.

Now that you're starting to think about research, I'd suggest that you ponder
the difference between research and development. PhDs are about research, not
development. Development is extending a programming language like Prolog to add
a new capability. Research is defining that capability and explaining why it
would be worth extending Prolog to add it. Yes, you might then actually extend
Prolog as part of your research, perhaps as a proof-of-concept justifying the
practical value of adding the capability, but the research horse must come
before the development cart. In academic research, you can live without the
cart, but you can't live without the horse.

Best of luck with your studies,

Randy

--
Randy Crawford http://www.ruf.rice.edu/~rand rand AT rice DOT edu

 

[Algernon]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Thanks for the answers. Yes, I assumed that some there would be some
coursework before research but in timetable, the student should be
publishing research or doing independent research during the first year if
the student wishes to receive any aid in the form of assistantship and in
order to stay in good standing in the program. I wanted to get a leg up by
having some ideas before I started classes.

Thanks again

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

"Algernon" <algernon@dmv.com> wrote in message news:<10dm8547d9m8ja6@corp.supernews.com>...
> I will be studying for my PhD starting this fall. The program stresses
> research. I have been trying to think of some areas and topics but am coming
> up blank. Does anybody have any ideas for some research topics; AI in
> general and Prolog in particular? And what are some periodicals,
> institutions that publish research papers.

There are several research groups that use logical methods. Here is an
interesting research group using action languages and answer sets for
reasoning, at UTA:

http://www.cs.utexas.edu/users/vl/tag/

I think you would also be interested in knowledge representation,
perhaps as applied to Semantic Web; that's about the only big area I
can think of where people want to deploy those big ontologies. (Half
seriously, of course, computational ontology is now a basic tool in
scientific computing) I can't tell how much AI Semantic Web really is,
though.

There is an advanced language that fuses logical and functional
languages, it's called mercury. Perhaps you would be interested in
semantics of new generation logic programming languages? There is much
to do in that area, our PLs are nowhere near perfect.

An ex-faculty of bilkent CS faculty, Pierre Flener, was interested in
automatic logic programming. Conceptually, that is *very* close to AI.
If you have a powerful automatic programmer, think of the AI problems
you can solve!

I came across one of his articles at a relevant AAAI report:

http://www.aaai.org/Press/Reports/Symposia/Spring/ss-02-05.html

I think the report would be a nice reading for a PhD candidate
interested in AI applications of Logic.

Cheers,

--
Eray Ozkural

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Eray Ozkural exa wrote:

> I think you would also be interested in knowledge representation,
> perhaps as applied to Semantic Web; that's about the only big area I
> can think of where people want to deploy those big ontologies. (Half
> seriously, of course, computational ontology is now a basic tool in
> scientific computing) I can't tell how much AI Semantic Web really is,
> though.
>

Yeah there is a lot of logic programming going on now with the Semantic
Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
Bnodes (unidentified descriptions) ... i think that is an area that is
ripe for AI ideas ... and there could be very tangible benefits to
getting it right.

[1] <http://infomesh.net/2001/cwm/>

patty

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

patty <pattyNO@SPAMicyberspace.net> wrote in message news:<%cyDc.103831$Hg2.64452@attbi_s04>...
> Eray Ozkural exa wrote:
>
> > I think you would also be interested in knowledge representation,
> > perhaps as applied to Semantic Web; that's about the only big area I
> > can think of where people want to deploy those big ontologies. (Half
> > seriously, of course, computational ontology is now a basic tool in
> > scientific computing) I can't tell how much AI Semantic Web really is,
> > though.
> >
>
> Yeah there is a lot of logic programming going on now with the Semantic
> Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
> Bnodes (unidentified descriptions) ... i think that is an area that is
> ripe for AI ideas ... and there could be very tangible benefits to
> getting it right.
>
> [1] <http://infomesh.net/2001/cwm/>

Ouch. Python! All that XML junk, it's gonna be slow....

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Eray Ozkural exa wrote:
> patty <pattyNO@SPAMicyberspace.net> wrote in message news:<%cyDc.103831$Hg2.64452@attbi_s04>...
>
>>Eray Ozkural exa wrote:
>>
>>
>>>I think you would also be interested in knowledge representation,
>>>perhaps as applied to Semantic Web; that's about the only big area I
>>>can think of where people want to deploy those big ontologies. (Half
>>>seriously, of course, computational ontology is now a basic tool in
>>>scientific computing) I can't tell how much AI Semantic Web really is,
>>>though.
>>>
>>
>>Yeah there is a lot of logic programming going on now with the Semantic
>>Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
>>Bnodes (unidentified descriptions) ... i think that is an area that is
>>ripe for AI ideas ... and there could be very tangible benefits to
>>getting it right.
>>
>>[1] <http://infomesh.net/2001/cwm/>
>
>
> Ouch. Python! All that XML junk, it's gonna be slow....

Yeah it's slow, but not because of python or all that XML junk. It's a
demonstration of something (mostly the utility of the N3 language and
making inferences) and were Tim to figure out what it should be doing,
it could well be speeded up. It *is* (well probably was) a fast way to
demonstrate and play with some concepts ... to see if they could be made
to work. I cited it here only as an instance of logic programming
happening on the Semantic Web. There are other, perhaps more practical,
examples, but those are not nearly as accessible.

The basic question, as i see it, is how do you even do reasoning with
the monotonic binary logic of OWL and RDF in the wild west of the
democratized web. Are there useful inferences to be computed? If so
where are the tools to get the job done?

patty

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

My opinion (for what it is worth) would be that there is far too much
noise to trust any inferred "knowledge" from the web.

Based upon my experience working with the Open Mind Commonsense
Database, which is structured, even if loosely, I think a trusted
source approach would be the only hope of valid results. The OM data
is full of a whole lot of garbage in addition to the valid attempts by
web donors, interspersed by probably well intended but nonetheless
violations of the structure.

Majority weighting could of course be applied, but I suspect any such
attempt would be subverted in similar fashion to the Google rating
games played by slashdotters.

Perhaps the best approach would be starting with trusted source sites
being granted exceptional weighting, followed by a majority weighted
skeptical review of unrated sources.

The editorial and review process in the Wiki world might be a good
indication of why majority review is needed in the web jungle...

Regards,

Ted Warring


patty <pattyNO@SPAMicyberspace.net> wrote in message news:<ZMWDc.122273$eu.2393@attbi_s02>...
> Eray Ozkural exa wrote:
> > patty <pattyNO@SPAMicyberspace.net> wrote in message news:<%cyDc.103831$Hg2.64452@attbi_s04>...
> >
> >>Eray Ozkural exa wrote:
> >>
> >>
> >>>I think you would also be interested in knowledge representation,
> >>>perhaps as applied to Semantic Web; that's about the only big area I
> >>>can think of where people want to deploy those big ontologies. (Half
> >>>seriously, of course, computational ontology is now a basic tool in
> >>>scientific computing) I can't tell how much AI Semantic Web really is,
> >>>though.
> >>>
> >>
> >>Yeah there is a lot of logic programming going on now with the Semantic
> >>Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
> >>Bnodes (unidentified descriptions) ... i think that is an area that is
> >>ripe for AI ideas ... and there could be very tangible benefits to
> >>getting it right.
> >>
> >>[1] <http://infomesh.net/2001/cwm/>
> >
> >
> > Ouch. Python! All that XML junk, it's gonna be slow....
>
> Yeah it's slow, but not because of python or all that XML junk. It's a
> demonstration of something (mostly the utility of the N3 language and
> making inferences) and were Tim to figure out what it should be doing,
> it could well be speeded up. It *is* (well probably was) a fast way to
> demonstrate and play with some concepts ... to see if they could be made
> to work. I cited it here only as an instance of logic programming
> happening on the Semantic Web. There are other, perhaps more practical,
> examples, but those are not nearly as accessible.
>
> The basic question, as i see it, is how do you even do reasoning with
> the monotonic binary logic of OWL and RDF in the wild west of the
> democratized web. Are there useful inferences to be computed? If so
> where are the tools to get the job done?
>
> patty

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

patty <pattyNO@SPAMicyberspace.net> wrote in message news:<ZMWDc.122273$eu.2393@attbi_s02>...
> Eray Ozkural exa wrote:
> > patty <pattyNO@SPAMicyberspace.net> wrote in message news:<%cyDc.103831$Hg2.64452@attbi_s04>...
> >
> >>Eray Ozkural exa wrote:
> >>
> >>
> >>>I think you would also be interested in knowledge representation,
> >>>perhaps as applied to Semantic Web; that's about the only big area I
> >>>can think of where people want to deploy those big ontologies. (Half
> >>>seriously, of course, computational ontology is now a basic tool in
> >>>scientific computing) I can't tell how much AI Semantic Web really is,
> >>>though.
> >>>
> >>
> >>Yeah there is a lot of logic programming going on now with the Semantic
> >>Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
> >>Bnodes (unidentified descriptions) ... i think that is an area that is
> >>ripe for AI ideas ... and there could be very tangible benefits to
> >>getting it right.
> >>
> >>[1] <http://infomesh.net/2001/cwm/>
> >
> >
> > Ouch. Python! All that XML junk, it's gonna be slow....
>
> Yeah it's slow, but not because of python or all that XML junk. It's a
> demonstration of something (mostly the utility of the N3 language and
> making inferences) and were Tim to figure out what it should be doing,
> it could well be speeded up. It *is* (well probably was) a fast way to
> demonstrate and play with some concepts ... to see if they could be made
> to work. I cited it here only as an instance of logic programming
> happening on the Semantic Web. There are other, perhaps more practical,
> examples, but those are not nearly as accessible.

I understand that it's a demonstration. I like Python as a scripting
language, and I think it's truly much better than visual basic,
tcl/tk, or perl. It has functional constructs, high level data
structures, etc. And its libraries I can only applaud; it's a
beautiful programming environment. Actually, I recommend python to
those just beginning to program, I think it's a balanced language.
However, I heard that for big projects it doesn't have a shining
performance... That was the gist of my remark. Likewise for XML. Even
in the fastest C++ processors, XML causes a lot of overhead. In KDE
project, XML UI framework shares a big part of the guilt in
application startup time.

> The basic question, as i see it, is how do you even do reasoning with
> the monotonic binary logic of OWL and RDF in the wild west of the
> democratized web. Are there useful inferences to be computed? If so
> where are the tools to get the job done?

My idea is that, humans are not likely to use Semantic Web tools at
all. I imagine that first we write some agents that compute the
semantics and then perhaps some useful inferences can be drawn... I
think that the "semantics" is not really contained in these ontology
languages, you need a programming language, not just some forward
chaining code.

Regards,

--
Eray Ozkural

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Eray Ozkural exa wrote:

> patty <pattyNO@SPAMicyberspace.net> wrote in message news:<ZMWDc.122273$eu.2393@attbi_s02>...
>
>>Eray Ozkural exa wrote:
>>
>>>patty <pattyNO@SPAMicyberspace.net> wrote in message news:<%cyDc.103831$Hg2.64452@attbi_s04>...
>>>
>>>
>>>>Eray Ozkural exa wrote:
>>>>
>>>>
>>>>
>>>>>I think you would also be interested in knowledge representation,
>>>>>perhaps as applied to Semantic Web; that's about the only big area I
>>>>>can think of where people want to deploy those big ontologies. (Half
>>>>>seriously, of course, computational ontology is now a basic tool in
>>>>>scientific computing) I can't tell how much AI Semantic Web really is,
>>>>>though.
>>>>>
>>>>
>>>>Yeah there is a lot of logic programming going on now with the Semantic
>>>>Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
>>>>Bnodes (unidentified descriptions) ... i think that is an area that is
>>>>ripe for AI ideas ... and there could be very tangible benefits to
>>>>getting it right.
>>>>
>>>>[1] <http://infomesh.net/2001/cwm/>
>>>
>>>
>>>Ouch. Python! All that XML junk, it's gonna be slow....
>>
>>Yeah it's slow, but not because of python or all that XML junk. It's a
>>demonstration of something (mostly the utility of the N3 language and
>>making inferences) and were Tim to figure out what it should be doing,
>>it could well be speeded up. It *is* (well probably was) a fast way to
>>demonstrate and play with some concepts ... to see if they could be made
>>to work. I cited it here only as an instance of logic programming
>>happening on the Semantic Web. There are other, perhaps more practical,
>>examples, but those are not nearly as accessible.
>
>
> I understand that it's a demonstration. I like Python as a scripting
> language, and I think it's truly much better than visual basic,
> tcl/tk, or perl. It has functional constructs, high level data
> structures, etc. And its libraries I can only applaud; it's a
> beautiful programming environment. Actually, I recommend python to
> those just beginning to program, I think it's a balanced language.
> However, I heard that for big projects it doesn't have a shining
> performance... That was the gist of my remark. Likewise for XML. Even
> in the fastest C++ processors, XML causes a lot of overhead. In KDE
> project, XML UI framework shares a big part of the guilt in
> application startup time.
>

Well i would not disagree with most of that. Strangely enough Tim's CWM
was a rebellion against XML giving us the streamlined N3 language. But
one wonders, as did Pat Hayes, why one does not just use KIF.
Incidentally the Semantic Web does not rest on XML, it rests on triples.
You can, if you are bold enough, use whatever surface language you want.


>
>>The basic question, as i see it, is how do you even do reasoning with
>>the monotonic binary logic of OWL and RDF in the wild west of the
>>democratized web. Are there useful inferences to be computed? If so
>>where are the tools to get the job done?
>
>
> My idea is that, humans are not likely to use Semantic Web tools at
> all. I imagine that first we write some agents that compute the
> semantics and then perhaps some useful inferences can be drawn... I
> think that the "semantics" is not really contained in these ontology
> languages, you need a programming language, not just some forward
> chaining code.
>

Well the semantics of a language are outside of the language. The
semantics is manifested by the game of the usage of the language. I
would think that would apply to programming languages as well.

Where it comes to the semantic web, i've always felt that the graph
*was* the semantics but apparently that idea didn't get legs in the W3C
working group. When you say "some agents that compute the semantics",
if the semantics are the graph, than that works because certainly an
agent can compute the graph. So i will go out on a limb and say "if the
semantic web works, it will be because the semantics are the graph",
Pat's lengthy tomb notwithstanding. Of course the trouble with that is
the graph is *not* supposed to be context sensitive - that is because of
its grounding in URIs which are not supposed to be context sensitive ...

.... look don't get me started ... ok .. ok ...

patty

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

patty <pattyNO@SPAMicyberspace.net> wrote in message news:<yEgEc.135978$Sw.40005@attbi_s51>...
>
> Well the semantics of a language are outside of the language. The
> semantics is manifested by the game of the usage of the language. I
> would think that would apply to programming languages as well.

To be accurate, that's called pragmatics, and no semantics of a
programming language is defined by the designer...

> Where it comes to the semantic web, i've always felt that the graph
> *was* the semantics but apparently that idea didn't get legs in the W3C
> working group. When you say "some agents that compute the semantics",
> if the semantics are the graph, than that works because certainly an
> agent can compute the graph. So i will go out on a limb and say "if the
> semantic web works, it will be because the semantics are the graph",
> Pat's lengthy tomb notwithstanding. Of course the trouble with that is
> the graph is *not* supposed to be context sensitive - that is because of
> its grounding in URIs which are not supposed to be context sensitive ...

I don't mean merging of the ontologies or the link graphs. Those are
easy to compute. I mean a web agent browsing the web, understanding
the stuff, and building meta-data appropriately (and selectively)...

I think those ontologies could work for specific domains like genetics
research, because their stuff has to be formal... They have a project
here with all the graphical ontology editors and what not. But I think
a political text or an artists' hang out is not like that...

It invokes a silly feel in me. It's like UML, mostly a burden on those
who need it least. Again, why on earth should a web page author be
concerned with a silly meta-data language at all?

> ... look don't get me started ... ok .. ok ...

Cheers,

--
Eray

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Ted Warring wrote:
> My opinion (for what it is worth) would be that there is far too much
> noise to trust any inferred "knowledge" from the web.
>
> Based upon my experience working with the Open Mind Commonsense
> Database, which is structured, even if loosely, I think a trusted
> source approach would be the only hope of valid results. The OM data
> is full of a whole lot of garbage in addition to the valid attempts by
> web donors, interspersed by probably well intended but nonetheless
> violations of the structure.
>

Yeah some of my own flakiness is perhaps still in OM. Did anyone ever
do anything useful with the data base ?


> Majority weighting could of course be applied, but I suspect any such
> attempt would be subverted in similar fashion to the Google rating
> games played by slashdotters.
>
> Perhaps the best approach would be starting with trusted source sites
> being granted exceptional weighting, followed by a majority weighted
> skeptical review of unrated sources.
>
> The editorial and review process in the Wiki world might be a good
> indication of why majority review is needed in the web jungle...
>

Well yes we really do need to get that web of trust going ... err ... or
... I'm thinking that a more biological approach might be called for.
I mean spam and viruses may just be the stimulus we need to get a
working system going. Think about it .. for every virus there is an
antibody.

just a thought ....

patty



> Regards,
>
> Ted Warring
>
>
> patty <pattyNO@SPAMicyberspace.net> wrote in message news:<ZMWDc.122273$eu.2393@attbi_s02>...
>
>>Eray Ozkural exa wrote:
>>
>>>patty <pattyNO@SPAMicyberspace.net> wrote in message news:<%cyDc.103831$Hg2.64452@attbi_s04>...
>>>
>>>
>>>>Eray Ozkural exa wrote:
>>>>
>>>>
>>>>
>>>>>I think you would also be interested in knowledge representation,
>>>>>perhaps as applied to Semantic Web; that's about the only big area I
>>>>>can think of where people want to deploy those big ontologies. (Half
>>>>>seriously, of course, computational ontology is now a basic tool in
>>>>>scientific computing) I can't tell how much AI Semantic Web really is,
>>>>>though.
>>>>>
>>>>
>>>>Yeah there is a lot of logic programming going on now with the Semantic
>>>>Web - see Tim Berners Lee's CWM [1]. One need is fuzzy matching of RDF
>>>>Bnodes (unidentified descriptions) ... i think that is an area that is
>>>>ripe for AI ideas ... and there could be very tangible benefits to
>>>>getting it right.
>>>>
>>>>[1] <http://infomesh.net/2001/cwm/>
>>>
>>>
>>>Ouch. Python! All that XML junk, it's gonna be slow....
>>
>>Yeah it's slow, but not because of python or all that XML junk. It's a
>>demonstration of something (mostly the utility of the N3 language and
>>making inferences) and were Tim to figure out what it should be doing,
>>it could well be speeded up. It *is* (well probably was) a fast way to
>>demonstrate and play with some concepts ... to see if they could be made
>>to work. I cited it here only as an instance of logic programming
>>happening on the Semantic Web. There are other, perhaps more practical,
>>examples, but those are not nearly as accessible.
>>
>>The basic question, as i see it, is how do you even do reasoning with
>>the monotonic binary logic of OWL and RDF in the wild west of the
>>democratized web. Are there useful inferences to be computed? If so
>>where are the tools to get the job done?
>>
>>patty

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

On 29 Jun 2004 16:18:03 -0700, erayo@bilkent.edu.tr (Eray Ozkural
exa) wrote:
>> agent can compute the graph. So i will go out on a limb and say "if the
>> semantic web works, it will be because the semantics are the graph",
>> Pat's lengthy tomb notwithstanding.

Er, "tome"?

>It invokes a silly feel in me. It's like UML, mostly a burden on those
>who need it least. Again, why on earth should a web page author be
>concerned with a silly meta-data language at all?

A couple of points. First, that a browser with ontology is free to
"ontologically" browse the raw page text without consideration of
explicit metadata on that page. Second, even old-style web pages have
various keyword metadata fields. Third, old-style or semantic-style,
various levels and flavors of explicit metadata supply another degree
of freedom, which can always seem to bring advantages in some cases,
but always at some cost in redundancy, increased effort, issues of
standard versus clashing ontologies, much less multiple
interpretations, etc.

Bottom line, I agree that the hand-coding of semantic data into the
text of a page is probably an ill-advised strategy. Just include
whatever magic keywords you want to use in the text itself, or in
visible headers, annotations, etc. Keep the smarts in the reader, not
the text. Learn how to write good. Er, well. That is, better.

J.

 

[Algernon]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Wow, Thanks for all the feedback. I usually dont get any replies to my
queries, now it seems that a full fledged discussion has been spawned. I am
going to look into the knowledge representation and ontologies. I am loath
however to have anything to do with the web. It seems to me that every
couple of years TBL comes up with something new, each time it comes with
promises of the new world order in distributed information. I read what he
said about RDF and it sounded like the same thing everyone said about XML,
and SGML before that and HTML before that. I mean nobody even followed the
standard for HTML so why should they for RDF. It seems to me that until
there is one standard (one that is actively enforced) for the representation
of knowledge that the semantic web won't get past syntax (pun intended).
Anyway is there any web pages about these triples. Anytime I search for
ontologies and such I come to RDF and XML.

Thanks Again

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

Algernon wrote:

> Wow, Thanks for all the feedback. I usually dont get any replies to my
> queries, now it seems that a full fledged discussion has been spawned. I am
> going to look into the knowledge representation and ontologies. I am loath
> however to have anything to do with the web.


You are making a big mistake. That is where the AI action is, and that
is where it should be. Why else would you find a Peter Norvig as the
director of search quality of the Google engineering department ?


> It seems to me that every
> couple of years TBL comes up with something new, each time it comes with
> promises of the new world order in distributed information.

He has been working on the same protocol stack for quite some time, in
fact it probably seems to him to be an inevitable outgrowth of the
information web he invented.

> I read what he
> said about RDF and it sounded like the same thing everyone said about XML,
> and SGML before that and HTML before that. I mean nobody even followed the
> standard for HTML so why should they for RDF. It seems to me that until
> there is one standard (one that is actively enforced) for the representation
> of knowledge that the semantic web won't get past syntax (pun intended).

Syntactically RDF is a recommended standard of the W3C. The IEEE has
been arguing over a standard upper level ontology now for some time.
Choosing that such standards are even useful is a topic of much
contention; and getting engineers to contribute and use such standards
is even more difficult. But that there can *be* one standard for
representation of knowledge in a context as wide as the Internet is at
best the idea of a neophyte.

> Anyway is there any web pages about these triples. Anytime I search for
> ontologies and such I come to RDF and XML.

Try the context
<http://www.google.com/search?q=rdf+triples>
or just read for it in the Concepts and Abstract Syntax document:
<http://www.w3.org/TR/2004/REC-rdf-concepts-20040210/>

patty

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

patty <pattyNO@SPAMicyberspace.net> wrote in message news:<VxWEc.7595>
> Except that wasn't me ... things are not always as they appear.
>

You mean there is more than 1 patty on the internet? I guess this web
thing really has become popular... 😛

>
> I think the system suffers by not adequately motivating people to
> contribute. One system that has become surprisingly useful is
> Wikipedia. Now i realize that the systems function in totally different
> domains; but both systems rely on unpaid contributions, and the one is
> motivated and the other is not.
>

So the trick then is to find some way of providing an ego reward for
participation...? Maybe some kind of ranking system showing the "top
contributors" or something?

> The trouble with a single-cell organism is that it carries its behavior
> with it. As such, most system administrators will view it as
> malevolent and squash it. Such a species has little chance of survival
> in todays Internet.
>

Yes, just so. Every the benefit approach will have to be balanced
with the potential for malware it brings. Very annoying.

Regards,

Ted Warring

 

[Guest]

Archived from groups: comp.ai.edu,comp.ai.games,comp.ai.nat-lang,comp.ai.philosophy,comp.lang.prolog (More info?)

"Algernon" <algernon@dmv.com> wrote in
news:10dm8547d9m8ja6@corp.supernews.com:

> I will be studying for my PhD starting this fall. The program stresses
> research. I have been trying to think of some areas and topics but am
> coming up blank. Does anybody have any ideas for some research topics;
> AI in general and Prolog in particular? And what are some periodicals,
> institutions that publish research papers.

The satisfiability problem is fun, if only tangentially related to Prolog.
In that, you'd probably want to concentrate on techniques for solving
instances of a particular structure (i.e., that encode a particular kind of
problem). See www.satlive.org for a good source of satisfiability papers
and other information. Plenty of stuff to research with that one.

> Thanks