factes, tags

cc. all@ Oscom
     +drupal core (pls copy locally)

Andrea R

(I am one of your electorate - ever so glad that you are on board)

I am currently migrating content-wire into drupal - it's a tough but meaningful exercise

We  just finished parsing the content, thru an autotagging system and ended up
with a massive, hiroshima type tag cloud explosion.

 Alphabetical sort is not enough, as already discussed

I see that you are doing some work in this area

http://www.facetag.org/

Please let me and the community have your suggestions as to what are the options
as we stand. Seek to implement this week :-)

Greatly  appreciated

Paola DM


_______________________________________________
General mailing list
General@...
http://oscom.org/cgi-bin/mailman/listinfo/general

[Andrea Resmini]

Paola,

> We  just finished parsing the content, thru an autotagging system and
> ended up with a massive, hiroshima type tag cloud explosion.

Yup. I can picture that easily. Gosh.

> I see that you are doing some work in this area
> http://www.facetag.org/

> Please let me and the community have your suggestions as to what are the
> options as we stand. Seek to implement this week :-)

That tag clouds have ux and usability limits and that we have to look to
alternatives / better ways to empower the users is now common knowledge.
There is a number of papers on this very subject, a lot of researching
and a number of web applications that are already trying out remedies
(say, RawSugar).

The problem with facets + Drupal, as far as I've investigated some
months ago, is that you can actually use the Taxonomy module as your
skeleton key (*) for most of the low-level work, but you currently lack
the ability to show faceted or clustered tag in a sensible and useful
way. Obviously, you cannot build a proper zooming interface as seen on
facetmap.com or in FaceTag either (**), and that is quite a issue in
terms of data to juggle between HTTP calls (you have to keep track of
where you are all of the time).

Generally speaking, the road to go for tag-laden web sites is faceting
or anyway 'splitting up tags in different meaningful categories', since
facets have a very precise scientific meaning as far as library sciences
are concerned, but right now, I do not think there is any out-of-the-box
solution: you could probably use a wise combination of vocabularies and
taxonomy modules to cluster tags, but I do not think this can be an
effective solution in the long run.

Furthermore, the issue is vast and I do not know anything specific about
your project, and that makes it quite hard for me to actually come out
with anything significative. Is that tagging 'closed' (let's say like in
a controlled vocabulary) or is it open to contributions (be it social or
editorial) and growing? That would make a difference in the way you can
anticipate (or can't) some of the outcomes. Why did you resort to
'tagging' as opposed to traditional IA in the first place?

As a sidenote, we showed a working prototype of the FaceTag engine at
EuroIA some weeks ago, and I had a chance to chat with Gunnar Langemark
(http://drupal.org/user/1075) about the fact that when I started to code
the thing I actually considered to code it as a Drupal module. I
ultimately let it go because that required a thorough comprehension of
Drupal's innards I didn't have time to build at the time. I started
toying around again with the idea after talking to Gunnar, so I'm far
from having anything even remotely intelligible or shareable, but I
thought then and think now that this could be a very interesting path to
follow (and I'm also confident that Drupal would probably make some of
the low-level gigs a breeze to code).

If you think that getting into a more detailed view of FaceTag's own
faceting engine would help or that you could benefit from a IA
brainstorming on these issues or even actual code, please let me know. I
think anyway that an IRC / Skype meeting would be more effective, then.

Keep me posted,

--

Andrea Resmini
vector@...


(*) I wrote a short sketch about that here:
http://www.resmini.net/columns/tabspace/implementing_facets_in_drupal

(**) I included a couple of screenshots in a private message to Paola,
not to clog the ML. Screenshots and slides from EuroIA and everything
should go online at www.facetag.org as soon as we have some time to
clean things up. Anybody interested, just drop me a line.


_______________________________________________
General mailing list
General@...
http://oscom.org/cgi-bin/mailman/listinfo/general

Thanks a lot Andrea
my first thought was 'we need to categorize the tags' but this was dismissed as a step backwards.

That was before we had a clear count. Now I think categores can simple be meta tags, or clusters as you say.
 I am looking for a semantic option of clustering the tags, thinking of a semantic network , except that I need a system to automatically make the inferences  and I can't think of where that exists, knoweldge neworks tend to be domain based while our site has a variety of conten

The tags, for your info, are currently just generated based on text analysis - and we simply have 5 years of innovative content. We did not want to use the old categories cause they were not good (developed inorganically, messy, etc) I am looking for a new categorizaton structure
I expect that I can purge at least 20o% of not primarily relevant tags, on the other hand I consider this tags somehow an asset that should be leveraged (not sure how)

I am now thinking how to handle the situation, and yes, lets perhaps think of developing a collaborative project. I am sure there will be drupal developers happy to tak up the challenge, but this is a universal problem and it would be very nice if we could end up with a totally interoperable module that can be easily plugged into different systems

I am sure there are a few possible approaches
including reparsing using selective criteria if necessary
(or whatever)
We are going to do some thinking over the next few days, we'll keep the list posted of progress
and yes, we'll skype you soon
thanks again
PDM

On 10/17/06, Andrea Resmini <vector@...> wrote:

Paola,

> We  just finished parsing the content, thru an autotagging system and
> ended up with a massive, hiroshima type tag cloud explosion.

Yup. I can picture that easily. Gosh.

> I see that you are doing some work in this area
> http://www.facetag.org/

> Please let me and the community have your suggestions as to what are the
> options as we stand. Seek to implement this week :-)

That tag clouds have ux and usability limits and that we have to look to
alternatives / better ways to empower the users is now common knowledge.
There is a number of papers on this very subject, a lot of researching
and a number of web applications that are already trying out remedies
(say, RawSugar).

The problem with facets + Drupal, as far as I've investigated some
months ago, is that you can actually use the Taxonomy module as your
skeleton key (*) for most of the low-level work, but you currently lack
the ability to show faceted or clustered tag in a sensible and useful
way. Obviously, you cannot build a proper zooming interface as seen on
facetmap.com or in FaceTag either (**), and that is quite a issue in
terms of data to juggle between HTTP calls (you have to keep track of
where you are all of the time).

Generally speaking, the road to go for tag-laden web sites is faceting
or anyway 'splitting up tags in different meaningful categories', since
facets have a very precise scientific meaning as far as library sciences
are concerned, but right now, I do not think there is any out-of-the-box
solution: you could probably use a wise combination of vocabularies and
taxonomy modules to cluster tags, but I do not think this can be an
effective solution in the long run.

Furthermore, the issue is vast and I do not know anything specific about
your project, and that makes it quite hard for me to actually come out
with anything significative. Is that tagging 'closed' (let's say like in
a controlled vocabulary) or is it open to contributions (be it social or
editorial) and growing? That would make a difference in the way you can
anticipate (or can't) some of the outcomes. Why did you resort to
'tagging' as opposed to traditional IA in the first place?

As a sidenote, we showed a working prototype of the FaceTag engine at
EuroIA some weeks ago, and I had a chance to chat with Gunnar Langemark
(http://drupal.org/user/1075) about the fact that when I started to code
the thing I actually considered to code it as a Drupal module. I
ultimately let it go because that required a thorough comprehension of
Drupal's innards I didn't have time to build at the time. I started
toying around again with the idea after talking to Gunnar, so I'm far
from having anything even remotely intelligible or shareable, but I
thought then and think now that this could be a very interesting path to
follow (and I'm also confident that Drupal would probably make some of
the low-level gigs a breeze to code).

If you think that getting into a more detailed view of FaceTag's own
faceting engine would help or that you could benefit from a IA
brainstorming on these issues or even actual code, please let me know. I
think anyway that an IRC / Skype meeting would be more effective, then.

Keep me posted,

--

Andrea Resmini
vector@...


(*) I wrote a short sketch about that here:
http://www.resmini.net/columns/tabspace/implementing_facets_in_drupal

(**) I included a couple of screenshots in a private message to Paola,
not to clog the ML. Screenshots and slides from EuroIA and everything
should go online at www.facetag.org as soon as we have some time to
clean things up. Anybody interested, just drop me a line.

_______________________________________________
General mailing list
General@...
http://oscom.org/cgi-bin/mailman/listinfo/general