On Wed, 09 Apr 2003, Yang Shouxun wrote:
> My training data contain statistical values for word combinations (or
> collocations) extracted from a corpus. The number is indeed very large.

Funny, I am currently also applying my tool to NLP (natural language
processing): because of the isomorphism between context-free grammars and
algebraic datatyes, it is possible to learn propositions about derivation
trees (or even more general: learn non-recursive functions). The problem
there is rather the size of CFG extracted from a large, annotated
corpus for German (many, many thousands of productions), which really
looks messy.

> I've learned this style in Scheme. Yet I feel paralyzed when trying to write 
> in it to build trees. The type declaration may make my point clearer.
> --8<--
> type  dtree = Dnode of dnode | Dtree of (dnode * int * dtree list)
> --8<--
> The problems are that unless the next call returns, the tree is not complete 
> yet and it may have several calls on itself.

But that's what the closure is for: it abstracts away the subtree that
still needs to be computed.

> I'm running Debian unstale. I checked just now on my laptop and "ulimit -s" 
> reurned "unlimited". I suppose the desktop that actually ran the program was 
> similarly configured.

Given that you already run into problems for comparatively small sizes,
I suppose that you are using the byte-code interpreter? Its builtin
stack space is 256KB, i.e. 64K-words.

> I also downloaded your AIFAD and had a cursive look at it. I found it
> does not handle continuous attributes yet and your design goal is quite
> different from mine. So I wrote mine from scratch and called it DTLR
> (Decision Tree Learner for Retrieval).

Yes, I haven't yet implemented handling of continuous attributes, because
I am aiming at an even more general system, where you can specify abstract
algebras (signatures) that describe how to handle values of some abstract
types, i.e. not only continuous (numeric) values. I have already done
so separately in another project, but wasn't very satisfied with the
design. Furthermore, I'd like to integrate it into AIFAD.

> If you are interested, I can send a copy to you tomorrow. It does not 
> implement all the features I planned, without documentation except some 
> comments, but it is enough for my own needs right now.

That would be great! - Thanks!

Regards,
Markus

-- 
Markus Mottl          http://www.oefai.at/~markus          markus@oefai.at

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners