Re : Are we losing out because of grammars? (Re: Schema ambiguit ydetection algorithm for RELAX (1/4))

["Vegt, Jan" <Jan.Vegt@softwareag.com>]

>Introduction to Algorithmic Information Theory, Nick Szabo; 
>http://www.best.com/~szabo/kolmogorov.html
>My thanks to Jan Vegt for reminding me about kolmogorov complexity
measures.
>These are more useful than Shannon's random source measures for our
purposes.

Len, 

Thanks you're a gentleman. I still like the Kolmogorov one-liner "if an
object contains regularities then it has a shorter description than itself".

In the mean time I've been trying to connect that to more practical levels
in supporting Rick's "two levels of grammar", basically I agree with Rick.
[ Allthough it should be broader than grammar defined as a set of rules ;
  it should also be applicable to say the set of all UK placenames etc. ]

I've just received Abiteboul's [et al.] "Data on the Web : from Relations to
Semistructured Data and XML".
http://www.amazon.com/exec/obidos/ASIN/155860622X/qid=980969273/sr=2-1/ref=s
c_b_1/103-8205445-0272624
It's a very readable and exciting book. Relevant here and now is that he
talks about 'schema extraction' defined as "given one particular data
instance, finding the most specific schema for it".

I know I bored you guys with the emphasisis on the differences between
structured and semi-structured data. But let me quote Abiteboul again "What
sets apart schemas for semistructured data from traditional schemas is the
fact that a given semistructured data instance can have more tha one schema.
This raises the following intriguing possibility : Given a semistructured
data instance for which we do not have any a priori knowledge, compute
automatically some schema for it;of course, given several possible answers,
we want the schema that best describes the structure of that particular
data. We call this problem schema extraction."

This sounds cool to me.

Later,

Jan

PS	Rick I may have some practical pointers for you later. Need some
time ...