Re: [Demexp-dev] Issues on classification

[felix . henry]

Hi,

I agree with you that a "simple" classification based solely on
keywords is far from ideal. After thinking, one main problem is that
a question can belong to several categories, which makes the parsing 
more difficult. Of course, in the S3 version (Stable, Scalable, Secure),
questions will belong to several categories, but the U2 version does
not need to have this feature. Also, as you mention, it would be nice
to be able to hierarchically parse the database in the GUI in U2 version.

Alternative possibilities:
-The user chooses the category (possibly creating a new one). We know that this
will no longer be feasible in a large scale system, however, smaller groups
wanting to use the DemExp independently may like this feature.
-The hierachy of categories is pre-set and/or evolved by a "person in charge"
and the user can only put his/her question in an existing category

Would this make the implementation easier?

I like the U2 name. I guess the ultimate version can be named T4 (Too much
Trouble To Think about).

   Felix



> Hello,
> 
> I have updated on CVS some few explanations on classification. However,
> the more I think about it, the more I think the solution I designed for
> keyword handling is Bad(tm).
> 
> To please F�lix, I think I will try to write a quick and dirty
> classification module that handle keywords as simple strings, without
> any handling of hierarchy (or just a minimal hierarchy, path matching)
> or classification rewrite.
> 
> The remaning modules to write are thus:
> 
>  - classification (only simple keywords and path matching)
> 
>  - server (taking commands from network and executing it). To accelerate
>    the development, the server will be single-threaded (poor performance
>    but easy to design, no locking issues)
> 
>  - network messaging (the minimal required set)
> 
>  - integrate Isabelle's work and make it emit network messages
> 
> All in one, a lot of work before our Initial Demonstration codenamed U2
> (Unsecure & Unscalable) release (do you see a better keyword for it?).
> 
> Any comment, idea, opinion?
> 
> Yours,
> d.
> -- 
>  David Mentr� <address@hidden>
>    http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
>  GnuPG key fingerprint: A7CD 7357 3EC4 1163 745B  7FD3 FB3E AD7C 2A18 BE9E
> 
> 
> _______________________________________________
> Demexp-dev mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/demexp-dev
>