tags – EdtechPost

thagoo – meta-search for social bookmark sites

http://www.thagoo.com/

So this kind of seems like a good idea, a site which allows you to search tags across a large number of social bookmarking sites, but somehow I expect I’ll use it as much as I use “meta-search” engines, which is to say, not at all.

If I knew I was getting the best stuff from all the sites, and it dealt well with duplication, then maybe. But my experience, which seems borne out by how hard it is to get people to shift off of their habitual search techniques, is that familiarity with the search experience (both interface and reliability of the results, and in the case of social bookmarking, the extent of your affinity or ability to control your sources) trumps the greater span of these ‘meta-‘ approaches. – SWL

Group-specific Tagclouds in Academic Portals

http://webtools.allegheny.edu/gnosh/

So the other gem for me from Bryan’s Educause article was the above Gnosh from Allegheny College. Actually, I’m not sure that it was this tool that excited so much as the idea it inspired.

One of the things that has always bugged me about broad tagclouds like the one on del.icio.us or flickr is that, well, they are really broad – there is nothing connecting all of the words appearing in the tagcloud other than that they were used by any user of one of these services, and the userbases on these services are totally heterogenous. So sure, I can see generally what the popular tags for all flickr or del.icio.us users are, but why should I care? What I do care about is what the tags used in my particular community are.

How many departments webpages or college portals provide search boxes to Google (or even their own sets of pages)? Lots, right? In both cases, the users using these search boxes have lots more in common than the entire set of users across flickr or del.icio.us, and in fact in cases like portals we typically can get real specific about group memberships and affinities. So if instead of passing their searches directly through to the search engines we first capture locally what the terms they were using, all of a sudden we can build tagclouds of search terms that are locally relevant to that community of users.

So as someone in the Faculty of Science taking this specific Biology course, I might come to my campus portal and beside my personalized search box see a tagcloud of terms that other people in the Faculty of Science had recently used (including my professors). Sort of like displaying the attention of my particular affinity group, and potentially opening up interesting terms I may not have thought to search on.

Probably not very 2.0’ish, and likely someone will scream ‘oh the invasion of privacy’ (though nobody is forcing you to use this search interface) but what I like about the idea is that it is imminently doable right now with almost no new tech – whether it be this gnosh piece or one of the other tag cloud pieces emerging out there, the only other piece would just be a small script that wrote the search term to a database table before passing it on to whatever search engine you were using (that could be configurable). Someone please tell me if this already all exists and I’m just being dense before I go off and spend the little time I have free to using my terrible development skills to hack this together, please….

(And as an afterthought – why can’t I see all of the tags used by myself and people I list as friends in flickr – or can I? how about all of the tags used by members of the same group? Maybe this is something that is already being done through the API?) – SWL

i d e a n t: Tag Literacy

http://ideant.typepad.com/ideant/2005/04/tag_literacy.html

I haven’t been doing a lot of ‘me too’ blogging of late (e.g. highlighting what other bloggers have written) but I thought this post deserved a mention, in part because I’m not sure if Ideant is as widely read as it should be. the piece is a worthwhile read for the folksonomies crowd. I like the term “distributed classification systems” – I’ve been using the term ‘dynamic taxonmies’ as my Furl category for such articles, but this term I thinks works better. I don’t have a lot of time for the term ‘folksonomies’ but at the end of the day, it’s hard to argue with a meme. – SWL

Distributed Tagging and Auto-Complete – an example

http://www.dwelle.org/avar.cgi

From a somewhat hysterical slashdot thread examining the user generated tagging systems in Flickr, del.icio.us and the like came a reference to this little experiment to introduce auto-completion and suggestion of del.icio.us tags based on a user’s previous tags. This is a step in the right direction – if it can start to pick up tags from the overall site, then maybe one of the issues with this overall approach, lack of synonym support and inconsistency applying tags, maybe isn’t as bad as it first appears. – SWL

Using Emergent Classification as a Starting (Not End) Point

http://www.adaptivepath.com/publications/essays/

From elearningpost comes mention of this useful article by Peter Merholz (some may remember him from ‘peterme‘ days, one of my earliest regular blog reads).

D’Arcy, King and I had been trading emails a few weeks back on the value of emergent classifications systems like those seen in Flickr for use in learning object repositories. Clearly, the idea is getting a bit of play, at least within the blogosphere.

What troubled me was that some of the current executions seemed a little bit like a baby/bathwater thing – yes, emergent classification systems are interesting and reflect actual users’ language usage, but they are also problematic – in being flattened, they do not have the depth (and the corresponding teaching ability) that hierarchical taxononmies can offer their users, and are also plagued with some of the problems Merholz points to. I mean, have you ever actually tried to find something you know should be there but didn’t know the classification for, (as opposed to just serendipidously browsing), in an flattened keyword system?

Instead, I think Merholz describes better than I did in my emails to D’Arcy and King what I think we should be looking towards – using ’emergent’ temrs as the basis for creating connections between terms users actually use, as the basis for continual refinement of more complicated, less flattened, taxonomies.

How would this actually work – at the very least I think it could show up in things like ‘type ahead’ functionality that tries to complete the term you are entering based on previous ’emergenet’ terms, or else asking the user to confirm whether they were using a term in one sense or another after they have submitted their choice. – SWL