Re: [Axiom-developer] topic modeling

[C Y]

--- root <address@hidden> wrote:

> I've muttered about expanding the latex tags to have \idea,
> \concept, etc. so that newly born papers could be more easily
> classified. However, a new technology seems to make that less
> interesting. It's called topic modeling. See
> 
> http://blogs.zdnet.com/emergingtech?p=304
> 
> Given a continuous process that scans incoming papers you
> could add them to a semantic network which models information
> that I find useful and in ways that I find useful.

Do you mean have each user of Axiom define their own model according to
their individual needs?  I suppose that makes sense to an extent.

A possible test for such a system might be to use the MSC2000 system -
some papers have subject assignment (or assignments) by their authors. 
A system for automatic classification could scan those papers, generate
primary and sub-classifications using MSC2000 categories, and then
check the generated results against those assigned by the authors. 
Eventually, journals might be able to automatically classify incoming
papers using such algorithms, although I think there will always be an
advantage to having a human evaluate the focus and application of a
paper.
 
> Axiom could strongly benefit from such technology if we could
> find a good source of technical papers.

I would think the only currently viable sources would be the major
journals in particular fields of interest, since they will carry the
major bodies of research for their fields.  Whether they would be
interested in such an idea I don't know.

> Starting from an initial
> source we could collect electronic conference papers and classify
> them. Then Axiom could just look up a concept like "Groebner Basis",
> follow it to "Homological Algebra" then onto "Computing P-modules"
> and then onto finding an algorithm for computing a presentation
> of a finitely generated P-module. Ideally the paper would be
> literate and contain code that was automatically incorporated
> into the system. 

An interesting idea indeed - unfortunately, the starting point would be
to retro-fit existing work into such a system, in order to have
something to build on.  The sheer magnitude of the task aside, I rather
doubt copyright, journal finanical self interest, and other such
mundane issues would allow such a broad program to take off in the
current climate, except possibly in a few sub-disiciplines.

> The Crystal idea has a facet that constantly watches the user
> interaction, maintians an "intensional stance" of the user,
> and tries to find related, relevant work. This technology would
> be ideal behind such a facet.

Indeed.

> Automatic classification algorithms are always more effective
> in limited domains (e.g. math) than in unspecified domains.
> 
> Sounds like an NSF or INRIA grant idea to me.
 
:-).

Cheers,
CY

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com