[Debtags-devel] Re: Recent progress

[Hervé Eychenne]

On Sun, Mar 06, 2005 at 11:08:30AM +0100, Benjamin Mesing wrote:

 Hi,

> > > My opinion is to make the vocabulary not too detailed. If you have
> > > something very specific in mind, for example a tool to manage your
> > > bookmarks choose e.g. "web::browser"
> > 
> > Wrong. If most of the Web browsers allow to manage bookmarks, you
> > could perfectly imagine a tool just for that (that is absolutely not
> > a web browser, I mean), as the task is complex enough to deserve some
> > application(s) of its own (check their validity, convert a bookmark
> > file format into another, associate... tags (!) to them, etc.).
> > But that was just an example.

> I didn't mean that the single application has to be a webbrowser but it
> is strongly related to a webbrowser and thereby it is natural to select
> the web::browser tag when searching for something to manage bookmarks.

Yes, but if you do that, you'll miss application that deal with
bookmarks and that are not a web brower. So it can be wrong.

> > > combine this search with a full
> > > text search for "bookmark" oder "manage" or even both and most like=
ly
> > > you will have what you are searching for (I just tried it, but curr=
ently
> > > this example does not work well, because neither gnobog,
> > > egroupware-bookmarks, or sitebar was tagged with web::browser -- I =
found
> > > those packages by performing a full text search for bookmark && man=
age).

> > > I know that a lot of you will disagree with a vocabulary being not =
to
> > > much detailed and combining it with a fulltext search.

> > For me, both things are orthogonal (and can be combined in a user
> > interface, then).

> And here I disagree, I think tags should be used for a rough selection
> and full text search can be used for the finetuning.

Two steps, then? Why not combine both in the user interface?

> That allows to keep the complexity of the vocabulary low.

Yes, but you don't really justify why it should be so.
I don't care about the complexity of backends. Backends can/must be as
powerful, complex, and expressive as possible. Only the user
interfaces may require some simplicity.

Tags are structured, and full text search is approximative.
Full text search is often unperfect (you cannot include every synonym of a
word in a package description). Tags are much more precise, that's why
I think we should concentrate on that, leaving the full text search
and other fuzzy logic things for upper layers (UI).

> > > The whole point
> > > of a vocabulary is to speak in common terms so to avoid inconsitent
> > > data. And I think you are right -- as long as professionals are inv=
olved
> > > this would be the way to go -- we need something with the focus on =
"easy
> > > to learn" in opposite to "easy to use".

> > Learn? But there should be nothing to learn... Users would use the
> > search tool from time to time, and we cannot expect them to have to l=
earn
> > something to use the search tool. So the user interface must be easy
> > to use (nothing to learn), and the tags are the backend, so it should
> > be as expressive as possible. That's my point of view.

> The user has to learn which tags are available

No. Not obligatorily.
If the user interface is well written enough, the user can enter
words, and the connexion with existing keywords should be done.

> - but you are correct
> this complexity can be reduced by a UI carfully designed. However you
> can't arbitarily simplify a complex system even with the most
> sophisticated UI. 

It's not about simplifying it, it's about giving a simple view.

> Searching a package can be split into two steps, first searching the
> correct tags and second searching the packages in the result.

Yes, but while it may be desirable to show the two steps explicitely,
they can be transparently combined.

> And the
> more complex the vocabulary will become the more time the first step
> will take - and the less the second step.

Isn't it equivalent in the end? ;-)

> So in the end it comes down to what we consider to be more important -
> exact search results or a vocabulary with a low complexity reducing the
> complexity of the first step and the tagging process.

> You vote for the first, me for the latter and I don't know who is
> right :-(

:-)
I think things must be as complex as they can be for people who can
afford dealing with power (backend), and as simple as possible for people
who don't care (UI). That's the UNIX philosophy, isn't it?

 Hervé

-- 
 _
(°=  Hervé Eychenne
//)  Homepage:          http://www.eychenne.org/
v_/_ WallFire project:  http://www.wallfire.org/