Wednesday, August 19, 2009

Can you use Zipf's law to determine what to digitize?

Ben Goldman, Digital Programs Archivist for the American Heritage Center at University of Wyoming, sent me a quick update from the Society of American Archivists Conference:

Mark Custer, digital librarian at East Carolina University "gave an amazing presentation on 'mass representation,' where he argued that it really isn't necessary for us to digitize *everything*. Using Zipf's Law and usage data from ECU's digital collections, he made a pretty compelling case that by digitizing representative portions of 20% of our collections, we could adequately serve 70% of our users."
I would be interested to know how much time it takes to do the analysis and also what users really think of it.

Technorati tag:


Archives Alive said...

This was a great presentation and unfortunately nobody recorded it. Even if he posts the slides, his talk along with it would be most useful. It was very different from the approach taken by UNC SHC, one of the other presentations in that session. I don't think analysis of users or success will be lost on either of these institutions.

Archives Alive said...

When I said nobody recorded it I was just kidding. You can buy a copy here:

Jill Hurst-Wahl said...

Unfortunately, that site doesn't tell you who the speakers were, so you need to go to the conference site in order to figure out which ones to purchase. Since I wasn't at the conference, I don't know if audio only would be good enough in order to understand what he said. For those that order it, I hope the answer is "yes".

kginfopro said...

Here's the name of the CD if anyone is interested: 601: Going With the Flow: Sustainable Models for Integrating Digitization . It's unfortunate that it isn't openly available.