Centuries ago, the best way to reproduce a book was to have a monk in a monastery sit down and rewrite the original, word for word.
These days, digitizing one of those ancient texts can seem almost as laborious: It can take hours upon hours of human work to scan just one volume. So George Washington University is now trying to figure out if an automated digitization system will take less time, and cost less per page, than a manual one.
The university announced Tuesday that it will use an automated system to digitize rare Middle Eastern texts from its own library and from that of Georgetown University. Library staffers will digitize hundreds of works over the next two years, and when the project is completed, they will examine the associated costs. They hope to be able to tell other libraries which method of digitization is more affordable.
Digitizing a book can involve disbinding it or having a human turn its pages. But at George Washington, a machine in the institution’s Melvin Gelman Library uses a black plastic arm to turn a page, pause as two cameras take pictures of both open pages, and then turn the page again. Air circulates through the arm of the machine, creating a gentle vacuum that can attract a page and guide it from the right side of the book to the left.
Lotfi Belkhir—the chief executive officer of Kirtas Technologies, the company that sold the machine to George Washington—showed guests at an event announcing the project’s start how the device works. The machine (above; image by William Atkins, George Washington U. photographer) whizzed through the yellowed pages of an 1899 translation of Omar Khayyam’s Rubaiyat. The machine, which costs $169,000, can photograph up to 3,000 pages in an hour.
Kirtas Technologies is based in the U.S., but most of its sales are to libraries in other countries, Mr. Belkhir said. George Washington’s purchase was subsidized by a grant from the Insitute for Museum and Library Services.
Buying an automated system is important because libraries need to determine the most cost-effective way to digitize their texts, said Karim Boughida, associate university librarian for digital initiatives and content management at the Gelman Library. The university wants to make cost information available to decision makers at other institutions, he said.
7 Responses to George Washington U. Experiments With Robotic Book Digitization
mbelvadi - December 2, 2009 at 11:55 am
This vacuum-arm technology isn’t that new. Last year, at a library conference I heard a presentation by a Canadian academic library digitization team that was using one of these (not sure of the exact model#, but same technology concept). The presenters admitted that they had all but abandoned using the vacuum-arm and were back to having students turn the pages by hand because the arm just didn’t work very well for their materials – they spent more time fixing errors than the time the students would spend turning pages. I hope that before GWU spent $170K on the machine that they actually talked to other libraries who were using it successfully, not just company sales reps, to see if the kinds/conditions of materials they plan to use it with will in fact be compatible with it.
ebookreader - December 2, 2009 at 5:56 pm
I for one excited to see that GWU is using the Kirtas technology to digitize their materials. I have actually used their technology quite successfully for more than 18 months. Their robotic arm uses vacuum to turn the pages, better than a human could ever accomplish. GWU is evaluating the technology, comparing it to manual technologies that can’t compare. I have used both, the Kirtas system was far superior, as we were able to scan books at almost 1600 pages per hour. I find it hard to believe that the arm had difficulty turning pages, the system can be adjusted on the fly, you can increase or decrease the vacuum, and they have blowers that puff up the pages, making them that much easier to turn. I for one think GWU will be very happy with their Kirtas Machine, the quality and production we were able to achieve with the machines I worked on was amazing.
johnwiley - December 2, 2009 at 6:55 pm
One extremely important aspect of the Kirtas machine is its gentle handling of the pages and the binding. I had some nineteenth-century volumes scanned and they came back exactly as I had sent them.
belkhir - December 2, 2009 at 7:50 pm
As the CEO and founder of Kirtas, my team and I are very honored to see our technology at the heart of this watershed project by GWU and Georgetown. We’re also pleased to have GWU be our first US customer for our next generation of highly reliable and efficient book digitization systems, i.e the KABIS family. It’s worth noting that our technology has been proven to work reliably and efficiently by hundreds of satisfied customers, including some of the most prestigious institutions around the world. Also our technology has digitized some of the rarest and most precious material automatically and with the highest quality. I invite mbelvadi and everyone else who reads this blog to check out our website at http://www.kirtasbooks.com to see for themselves some of the unique collections we’ve digitized with our technology, and that are now available for worldwide online access and print on demand. After you do that, please compare our digitized books with those produced by other well known book digitization projects. That is the bottom line…
boughida - December 3, 2009 at 10:17 am
I am the primary responsible of this project. Assuming that GWU would buy such equipment without doing an evaluation is extremely naive and simplistic. GWU and Georgetown staff members involved in this project have a wealthy of expertise in the domain of digitization including the manual scanning technology you are mentioning in your entry. Do you know the TCO (Total Cost of Ownership) of such “leased” manual equipment? do you know that you cannot buy them (as of two years ago)? Besides Kirtas digitization systems we have all kind of state of the art services that goes with it eg support for XML METS/ALTO, MIX, MODS, MARCXML, OCR, workflow management, etc…
bfrank1 - December 3, 2009 at 12:34 pm
Anyone who has ever stood in a press room listening to the chuff and wheeze of offset printers as they use “vacuum technology” to move papers through the guts of the press will not find this story very amazing. Modifying the nearly 100 year old printing methodologies to work on sheets bound on one side does not seem particularly heroic, but every paper has a different weight and finish, and having been up to my armpits in wadded up press disasters, it is no simple thing to keep “robots” of this type happy. As usual the dreams of automation are held hostage by the stubborn physical realities of objects.
allens - December 3, 2009 at 8:41 pm
Might I suggest that, if there are some skipped pages (which I’ve experienced with _manual_ scanning of articles at libraries!), that one way to handle this would be to scan each book twice, with the book flipped over the second time (and thus likely to be picked up in a different way by the robot arm)? Then run OCR on both versions and compare.