[Debtags-devel] Proposed Debtags goals for Etch

[Peter Rockai (mornfall)]

Hello folks,

On Monday 25 July 2005 09:58, Enrico Zini wrote:
[snip almost-flame]
>  - is Czech Republic western or eastern-europe?
central, goddamnit! ;-)
>
> It's easy to start a hierarchy, much less so to make it model the real
> world without forcing the real world to fit in the hierarchy.
>
> In some case, maybe with file formats, you can do this forcing.  In
> other cases, this forcing will result in imposing your point of view,
> stereotypes and preconceptions over other people.  This is a risk when
> doing categorization, and I'd like to do all I can to avoid it.

I think both of you have a valid point here. I _think_ what Erich suggest is 
to treat, say works-with::audio:* same like there was a works-with:audio:: 
facet (which is actually also the same as your approach with "mini-facets", 
just adds a naming convention). Actually, you two seem to be speaking of the 
same thing in different terms all the time. How i understand this:
If instance X of concept A is in itself a concept (B), you want to add 
instances of B too.

You could do that by making B = A:X, but there is a danger of existence of 
concept C and its instance Y, such that in fact C::Y represents same concept 
as B. Here, we get the conflict, where B = A:X = C:Y and install redundancy 
into our primary dataset. This is probably not wanted. We could shortcut this 
by allowing facet aliases, so say A:X = C:Y. We get a small facet-space 
explosion here though, so maybe don't do that.

Probably it would be better to really rename it to B. I am not familiar with 
enrico's tag and facet hiding algorithms, but i'd say that if facet B implies 
A::X || C::Y, we can as well hide it and show it when user picks one of those 
2 tags. B is not going to have many packages anyway. If it outgrows the 
implication, it'll get into the "more important" facet set. Now for this, we 
need for each facet a set of tags from _other_ facets that are implied by the 
presence of this facet, but i think enrico has that already? And 
progressively hide everything that is implies few tags till we get a 
reasonably small (7-8) facets to start with. As noted elsewhere, this could 
be aided by the fulltext search, so we come up with more relevant starting 
set.

</rambling>

>
>
> Ciao,
>
> Enrico
>
Yours,
    Peter.
-- 
Peter Rockai | me()mornfall!net | prockai()redhat!com | +421907533216 
   http://blog.mornfall.net | http://web.mornfall.net

"In My Egotistical Opinion, most people's C programs should be
 indented six feet downward and covered with dirt."
     -- Blair P. Houghton on the subject of C program indentation