[Zope] Something odd about ZCatalog...

Michael Bernstein mbernstein@profitscape.net
Mon, 13 Sep 1999 15:20:02 -0700


Michel Pelletier wrote:

> > However
> > I was disappointed at the time that all such vocabularies had
> > to be created
> > and maintained by hand, even if you were using an existing
> > one (say from an online
> > glossary),
>
> This is what I was thinking, what else did you have in mind?  Unless you
> can teach a computer to read and understand context and applicability of
> words, I don't see how it can be automated.

Umm, I meant that if a new document is added to the system that matches an
already defined keyword (ie: a word that is already part of the sites managed
vocabulary), no more needs to be done. When I say 'maintain by hand' I'm
refering to editing HTML files directly. Now, I realize that there are some very
heavy-duty document management solutions out there, but no-one is going to
spring for what those systems cost for a relatively small website, no matter how
productivity and usability are going to be improved.

Instead, most businesses opt to maintain this sort of thing by hand, with all
the attendant problems of linkrot.

Now when I say 'controlled vocabulary' I mean being able to do things like
search for 'car' and get all matches for occurences of 'automobile' as well. A
controlled vocabulary also gives you an immediate head start in defining a
consistent labeling scheme.

>  It might make sense to have
> a user interface where the manager can 'ask' the catalog if a word will
> be stopped or not, and be given the option to add or remove the question
> from the catalog, ala:
>
> Manager:  Is 'Many' a stop word?
>
> Catalog:  Yes, would you like to remove Many as a stopword?
>
> Manger:   Yes.  Is 'antidisestablishtarianism' a stopword?
>
> Catalog:  No, would you like me to add it as a stopword?
>
> Given this interactivity, it might be at least more useful than a flat
> list of words you can edit.

That sounds quite a bit more functional than what I had in mind, actually.

> > I've never had a client that would spring for
> > creating a vocabulary
> > maintenance tool. ('why do we need that?').
>
> That's reasuring.  Maybe since you understand how hard it is, *you* can
> spring for it?  ;)

This issue has come up before:
http://www.zope.org/pipermail/zope/1999-August/007593.html

I AM actually willing to put up some Ca$h to get my 'pet' ideas incorporated
into Zope. However, I would really like to have a more formal proccess for
aggregating demand of this sort from other 'friends of Zope' into a more
meaningful financial contribution. In effect, a Zope-specific SourceExchange
that would let non-coders pool resources to add particular features.

Digital Creations could also list projects waiting on the back burner, and let
aggregated financial contributions from the Zope community drive their
completion, rather than have them wait for one big client with a need and a
willingness to pay.

Other 'pet' ideas for inclusion in Zope:

- a 'pointer to object' property type for ZClasses (for many to many
relationships)
- a 'pseudo heirarchy' object (seems as though Topics is going to be this)
- 'naked' versions of the add/edit forms for ZClasses for direct inclusion in a
template

It would also be nice if this sort of financial contribution (ie: not a specific
client driven demand), would get the contributor something 'extra' in the way of
bennies.

> > With a controlled vocabulary for cataloging and retrieving
> > objects, and Topics to
> > arrange objects into arbitrary hierarchies, Zope could
> > credibly claim to have one of
> > the most advanced Web Content Management Systems on the market.
>
> Wait, I thought it was allready?

That depends. If you interpret 'Web Content Management System' to mean 'System
for management of content through the web' , well then, you're right. If you
interpret it as meaning 'System for management of web content' then not
necesarily.

Please understand that although I think Zope is the first credible 'all-in-one'
solution, there are many domain specific products that do a better job in a
single problem domain. So though I appreciate Zope's object-oriented app
building methodology for example, that doesn't necessarily give Zope a boost in
the content management problem space.

However, I feel you could easily claim 'Zope is the most advanced Content
Management System Toolkit', and this statement would be true even though I did
not include the 'web' qualifier.

Cheers,

Michael Bernstein.