CIL: Specialty engines Gary Price, Raul Valdes-Perez, Jefferey LaPlante

“what hasn’t changed much in seven years is how hard people are willing to work at searching” AP 10/04
Still only first page, 2.8 words per query, give up after 5 minutes

(GP)Commercial side realizes:
AKA vertical search, niche engines, usibility of data in a specialized interface

(JL) Xrefer
RUSA study – only 9% of in-library reference questions were answered with reference books
Students searching the web find more questions, not answers, and get a myopic view.
Book budgets are shrinking (ok, tell me something new)

43 publishers 170 titles – Cross searchable. Visualization tools.
~~~advertisement omitted~~~
Now he’s telling us we have to market library services. Duh.

(R V-P) Vivisimo
Bookstores vs. Electronic World
In a decent bookstore you don’t just see piles of things on the floor, things are in neat order on the shelf. In the electronic world, that’s not the case. You get a big pile. People have tried to fix this by applying rules from the print world – hierarchical listings. Doesn’t work.

He’s demonstrating clusty – they’ve added Teoma, dropped Overture. Works on the fly, spontaneous, helps you discover new ideas,
Tabs for channels of information – there’s a new government tab. Click on details to see how many results from each resource.

Of the competing ways – refinement, extraction, or clustering – they’ve gone in for clustering.

Pitt has licensed this for the Digital library– they use a meta-search but it just gives them a pile of 25 from each database, not de-duped not integrated. W/clusty it groups and orders the results

Question from the audience – what is clustering – in 200ms the computer looks at the text of the results, finds common concepts, groups those sites together, creates heading on the fly.

Laura (a LJ Mover and Shaker) pointed out something very interesting – how about a way to combine the folksonomies with the clustering? Maybe it could cluster the tags? Maybe the tags could inform the clustering – be weighted more heavily? Clusty does have a blog tab that searches Technorati—but I haven’t used it thinking about tags. Hmm. Have to come back to that.

Funny little note, they had a clip from me on their site as negative feedback for their search! How great is that.
Update: I verified that it was my quote -- but it was from a forum, not my blog. It's linked above.


CIL2005conferencessearch engines


  ¶ 1:45 PM| |cites (technorati) |