Re: [LogiLogi-list] On a simplification of tags: smart lattice

Hi Wybo,

I really like this aproach, we are injecting simplicity and
flexibility to LL and this is esential for usability, we are heading
the right way.

I'm already having some visions :) about the UI for this and I'd like
to ask some details... As from your link example, I see that there
will be giving the user the posibility to add either content or
context tags (in the old concept)... I think this should also be the
way to navigate through logis, and I'll make a graphical proposal
about this.

Back to the IRC now :)

greetings!

On Thu, Aug 28, 2008 at 7:51 AM, Wybo Wiersma <wy...@lo...> wrote:
>
> After another IRC-conversation with Bruno, I think we can make things
> even simpler, without losing important features & while gaining a lot
> of clarity...
>
> What if for the navigation we used a normal concept-lattice: that is
> have only one type of tags, while still giving logis a fixed tag-order,
> and telling the user the first tag is most important.
>
> First the concept lattice will:
> * Make the cloud fuller even with little content, as all tags are used
>  inside it
> * Make finding things easier, as Aristotle, History and History,
>  Aristotle would show up when browsing to History, Aristotle.
>
> Secondly the fixed tag-order for the logi's link will
> * Provide stabillity, and via links induce sensible tagging by default.
>
> This because when adding a link we could show 2 options to the user:
>
> * To a more specific sub-tag
>
> Aristotle , Philosophy , |_     | (user can key in Ethics for example)
>
> * Or to somewhere else
>
> |_     | , Philosophy (user can key in Sartre)
>
> (and there is a reset sub-tags button)
>
> Also when navigating there can be - behind the scenes - 2 searches.
> One for the exact tag-order "History, Aristotle", and one for the tags
> in any order, and with other sub-tags if needed...
>
> Results from this first search will then always be shown first, no
> matter their rating, followed by the other results.
>
> The advantage of this smart lattice-approach is that to the user it
> just are tags, plain and simple, and in the search-layer too, so we
> just need one cloud there, while the specialization of logis is still
> possible and easy, and their locations rather stable...
>
> What do you Bruno & others think of this ?
>
> greetings,
>
> Wybo
>
> > ---
> >
> > Besides that I have been thinking about the context-content links
> > stuff and about how complicated it all is to unsuspecting visitors, so
> > we (Bruno and I) 'd like to make it simpler...
> >
> > What I'd like to propose is first of all to - for now again - only
> > allow one tag-set per logi... Something like 'Aristotle, History, Life'
> >
> > This tag-set will be the fixed location of the logi, and still multiple
> > logis can have the same tag-set (that is there still are contenders).
> >
> > Also still some tags are more important than others, that is we still
> > have content- and context-tags, but for clarity we could call them the
> > primary tag, and context-tags.
> >
> > Users visiting LogiLogi will, on the mainpage see a cloud with all
> > primary tags, so at least Aristotle in this example.
> >
> > Then, they can click a primary tag, giving them the url /Aristotle.
> > They then will see a new, smaller cloud, below the url, on the right.
> > In this cloud the context-tags are shown (History and Life in this
> > example). Below the url on the left are snippets of all logis that
> > have Aristotle as their primary tag, sorted by their rating (for the
> > user's selected peergroup). Visitors can click on one immediately.
> >
> > When they like to narrow their search, they can also click one of the
> > context-tags in the smaller cloud (for example Life), and then a new
> > query is made, showing only the logis that have Aristotle as their
> > primary tag, and Life as one of their context-tags.
> >
> > The 2 assumptions behind this proposal are
> > - that people, when looking for a page start vague, but with the
> >   subject, and then narrow down accordingly.
> > - that it's good to keep things entity oriented; to draw different
> >   Logi's on each subject (like Aristotle) together, and to keep things
> >   easily navigatable... This should lead to interesting discussions.
> >
> > In this proposal there will still be requested and received url-bars
> > in the logi-view, but not in the search-view. In the logi-view (also
> > shown when there is only one logi for a set of searched for tags &
> > no logi's below it in the tree) the requested url-bar will contain the
> > tags of the query, while the received url-bar will contain the tags
> > used when the logi was created (also in the order then used). Some-
> > times this can be more (or less, when the user typed them directly)
> > tags than were in the query.
> >
> > This allows new people to - when creating new logis - put things in
> > the same or simmilar places as earlier logis, thereby making it easier
> > to keep things clean than not...
> >
> > To make this even stronger, also when adding links to a logi, by
> > default the context tags of the logi being added are already filled in
> > for the new link. So a link from Aristotle, History to Socrates, will,
> > without user-interaction be to Socrates, History. This; be it that it
> > is in reversed tag-order now, was something that really worked well in
> > the old version of LogiLogi, but now, with easy navigation it will
> > work even better!
> >
> > An additional advantage of this approach is that we can use Ferret (or
> > a better search lib) for most of the searching with this method, so it
> > will also scale very well... (the only moderately slow thing will be
> > the generation of the clouds).
> >
> > Fully implementing & testing this proposal in the back-end will take,
> > I think between 1 and 2 weeks, so not too bad either.
> >
> > I will do some more pondering about it, but I was just wondering what
> > you pplz (Bruno, Miguel & the others) think of it...
> >
> > Wybo
> >
> > PS: Yup it will be primary-tag first, instead of primary tag last, as
> > was currently the case in the url & in the old version of LogiLogi. I
> > was still thinking about it the other way when we chatted yesterday
> > Bruno, but I guess some things you said, changed my mind...
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> LogiLogi-list mailing list
> Log...@li...
> https://lists.sourceforge.net/lists/listinfo/logilogi-list



--
Bruno