[Top] [All Lists]

Re: [ontolog-forum] Kolmogorov Inductive Inference

To: "'[ontolog-forum] '" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Rich Cooper" <rich@xxxxxxxxxxxxxxxxxxxxxx>
Date: Fri, 2 Apr 2010 11:36:29 -0700
Message-id: <20100402183638.CDF1B138D3B@xxxxxxxxxxxxxxxxx>


Rich Cooper


Rich AT EnglishLogicKernel DOT com

9 4 9 \ 5 2 5 - 5 7 1 2


Hi John,


John Bottoms wrote:



SGML consists of a grammar for the _expression_ of grammars.

During the early days of the CALS initiative there was

discussion of tokenized version of a new DTD tag set. It was

called Binary SGML or BSGML. It didn't get much traction but

it allowed the reduction in size of the tags. One probably

doesn't want to mess with the data that the tags envelop. That

would fall under something like Markov models or some codebook



You can find scraps of it on forums such as this:



-John Bottoms

  FirstStar Networks

  Concord, MA 01742

  T: 978-505-9878


RGC:> Thanks John, that's an interesting tidbit of history I wasn't aware of.  The same thread took me to a discussion of "xml binary", which seems to have been a movement with small following, but clearly useful for embedded appliances.  Trading xml event messages seems to have made the inter-enterprise data sharing work rather well, even among different character sets and human languages.  The only application area that occurs to me for xml binary seems to be for embedded intranets with ultra fast response times (are there any such applications in public use?).  


But I think the main compression value for xml binary and sgml metalanguages is mostly in its linear compressive ability.  The Kolmogorov Complexity work centrally focuses on compressions that are more than linear, i.e., ability to substitute an arbitrarily complex (linguistic?) model for a common phrase as used in the data.  The KC invariance theorem in my earlier post, in which one language (i.e. one model of behavior in the data) MUST differ in size from another language model ONLY by a constant factor, is a VERY illuminating result.  


For example, it explains how the stack of symbolic frames we project into conversations among familiars can be modeled and represented in layers.  The notion of object-oriented context corresponds nicely to each frame in the stack.  The same notion might help formulate methods for developing compression models.  





Rich Cooper wrote:


> Hi Ontologizers,


> I've been comparing and contrasting Popper's works against Kolmogorov

> complexity ideas.  This is sure to be of interest to linguistically

> inclined ontologizers:



> This is a snippet from a tutorial on Kolmogorov Complexity which I found

> on the web somewhere and am reading next to Popper. 


> It seems to me that this compression-based model of data applies very

> well to linguistics ontologies, since language is compressed in so very

> many ways.  Curiously, I haven’t found any psychological studies

> relating this to language behaviors, or to any kind of ontological

> behaviors.  Does anyone know of such materials?  What are the jargon

> keywords needed to research that material by googling?  Who are the most

> fruitfully publishing authors?


> Suggestions appreciated,


> -Rich


> Sincerely,


> Rich Cooper

> EnglishLogicKernel.com

> Rich AT EnglishLogicKernel DOT com

> 9 4 9 \ 5 2 5 - 5 7 1 2

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (01)

<Prev in Thread] Current Thread [Next in Thread>