Google and Semantics
Locked
 
|
Re the discussion below, while Google doesn't, Clusty makes a basic stab.
See http://clusty.com/ > When I do a search with Google I see very little 'intelligence' of that > kind in the results. There appears to be some statistical weighting, > but the 'intelligence' of the results seems to depend entirely on > whether my word combination captures the concept I'm looking for. I > don't believe that's definable by any means I know of yet. > Yes. As far as I'm aware Google has not yet deployed a production quality technology for the semantic web. Google doesn't reason about concepts. Not only can't it trim down logically inappropriate results, it can't expand on related concepts unless there happens to be data (like from Wikipedia) where someone has created a document that physically contains the overlap of different nomenclatures. It certainly can't tell you whether two mathematical formulations of similar models will make the same predictions unless, again, there happens to be a web page posting of someone that said it was so. -- George T. Duncan Professor of Statistics Heinz School of Public Policy and Management Carnegie Mellon University Pittsburgh, PA 15213 (412) 268-2172 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://redfish.com/pipermail/friam_redfish.com/attachments/20070422/8fa16e04/attachment.html |
|
|
George Duncan wrote:
> Re the discussion below, while Google doesn't, Clusty makes a basic > stab. See http://clusty.com/ As I understand it, Clusty is doing cluster analysis in a statistical way, and does not represent things in relations of objects and actions, etc. It doesn't make or crawl RDF/OWL or model natural language. http://www.hakia.com aims to a `meaning' oriented search engine. When I searched for `global warming', Hakia gave me a breakdown of categories, including stuff like "Possible Solutions", "Movies and Documentaries", and so on. Also Hakia has a dialogue system where terminology can be clarified (as if one was working with a reference librarian). These pages illustrate and describe how queries are modeled: http://labs.hakia.com/OntoSem/hakia-lab-ontox.aspx http://labs.hakia.com/hakia-lab-onto.html http://www.ontologicalsemantics.com Also, I ran across this leak of Google's "Big Goals and Directions 2006". http://blog.outer-court.com/archive/2006-10-26-n80.html where it says: "Google wants to have the world?s top AI research laboratory." Also relevant are Larry Page's remarks here: http://technology.guardian.co.uk/news/story/0,,1781121,00.html "Mr Page said one thing that he had learned since Google launched eight years ago was that technology can change faster than expected, and that AI could be a reality within a few years" |
|
|
George Duncan wrote:
> Not bad at a "meaning" level, I think. Also useful for the searcher. > Clusty is a Carnegie Mellon spinoff from CS. A lot of the research > on information retrieval done here works with rather simple > (conceptually at least) statistical models. Here's a link with a broad > overview: http://www.lti.cscmu.edu/Research/index.html > <http://www.lti.cs.cmu.edu/Research/index.html> Thanks for the link -- looks like their machine translation and information retrieval projects follow both statistical and grammatical approaches. For web search engines, at least for casual users, I think its pretty clear that stateless clustering approaches can work well. My interest is whether, using automated procedures, scientific terms can be determined to have consistent meanings or not. If it didn't matter what order words and sentences had, words' part-of-speech, etc. then we ought to be able to scramble any text and still understand it. (How basic statistical retrieval systems work.) |
|
|
Actually, the context of other replies covered my questions on the how
computers can link concepts, but not on how people somehow have such separate languages that learning from one sphere can't cross over to the other. There's still that conceptual gap between promoting more efficient growth and a desire to limit the real economic impacts on the earth. All the communities I have contacted on the subject are speechless for some reason. The clear evidence of a gap in intellect is that global economic efficiency is methodically improving at half the rate of output for energy at least (Jevons Paradox, or the natural vanishing returns of efficiency as you may call it). You see the same pattern in the general technology cycle that demonstrates a growth and climax development model of terminally limiting efficiency for any physical process known, and for good reasons applying generally to any possible undiscovered one too, of the same kind as predicted by the 2nd law for energy transformation. It seems that the 1st law of human behavior, not to look at the 2nd law, still predominates even among scientists, but since the world is physically running smack into it at an accelerating rate, something will give. Phil Henshaw ????.?? ? `?.???? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 680 Ft. Washington Ave NY NY 10040 tel: 212-795-4844 e-mail: pfh at synapse9.com explorations: www.synapse9.com > -----Original Message----- > From: friam-bounces at redfish.com > [mailto:friam-bounces at redfish.com] On Behalf Of Marcus G. Daniels > Sent: Sunday, April 22, 2007 11:54 PM > To: The Friday Morning Applied Complexity Coffee Group > Subject: Re: [FRIAM] Google and Semantics > > > George Duncan wrote: > > Not bad at a "meaning" level, I think. Also useful for the searcher. > > Clusty is a Carnegie Mellon spinoff from CS. A lot of the research > > on information retrieval done here works with rather simple > > (conceptually at least) statistical models. Here's a link > with a broad > > overview: http://www.lti.cscmu.edu/Research/index.html > > <http://www.lti.cs.cmu.edu/Research/index.html> > Thanks for the link -- looks like their machine translation and > information retrieval projects follow both statistical and > grammatical > approaches. For web search engines, at least for casual users, I > think its pretty clear that stateless clustering approaches can work > well. My interest is whether, using automated procedures, > scientific > terms can be determined to have consistent meanings or not. If it > didn't matter what order words and sentences had, words' > part-of-speech, > etc. then we ought to be able to scramble any text and still > understand > it. (How basic statistical retrieval systems work.) > > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > lectures, archives, unsubscribe, maps at http://www.friam.org > > |
«
Return to Friam
|
1 view|%1 views
| Free forum by Nabble | Edit this page |