We use cookies to provide you with a better experience. If you continue to use this site, we'll assume you're happy with this. Alternatively, click here to find out how to manage these cookies

hide cookie message
80,258 News Articles

Google bringing search to historical manuscripts

Using shape-matching technology

History buffs can search George Washington's manuscripts online today for terms such as 'revolution', but only thanks to the tireless workers who transcribed the hand-written documents into digital form.

Soon, many other hand-written historical documents could be made available for the public to search - and through considerably less effort - if a research project funded by Google and being executed by three universities works out as planned.

The project, announced by DCU (Dublin City University) yesterday, started on a whim. DCU professor Alan Smeaton has been working on technology that can recognise objects that appear in videos. His technology can detect an object, such as a car or an airplane, in the frame of a video, then extract the image to compare it to a database of images to identify it or enable it to be searched.

Smeaton and his colleagues decided to find out if their shape-matching technology could be used to identify words, so they tried it out on the archive of former US President George Washington, which consists of 304,000 digital images and is available on the Library of Congress website. It worked well, Smeaton said.

Smeaton decided to use George Washington's archive because it includes hand written documents that have been transcribed. That meant that he could compare the results from his technology with the results from the current search system.

He had been talking to people he knows who work at Google in Dublin about the video-matching technology, and happened to mention the George Washington manuscript trial. "They were interested so we did some more experiments and showed them the results and they decided to fund a project," he said.

Smeaton wouldn't say how much funding Google has committed but said it will cover a year's worth of work by three or four researchers at DCU, as well as the same number of researchers each at the University of Buffalo and the University of Massachusetts at Amherst.

The goal of the project is to demonstrate that the technique is workable and scalable, Smeaton said. If so, Google can decide to employ the technology. The researchers are not locked into making the technology available only to Google, however, Smeaton said. They plan to publish their findings as scientific research.

Ironically, it's easier to apply the technology to some manuscripts that are much older than Washington's. DCU is also involved in a project with the Dublin Institute of Advanced Studies which is digitising manuscripts, the oldest of which dates back to the twelfth century, written in Irish. Those documents, beautifully and ornately designed by monks, are actually much easier to develop a search mechanism for, Smeaton said. "The monks were laboriously toiling over this and using great consistency across entire manuscripts," he said. "George Washington wouldn't be."

Google has also been at work scanning books from large libraries in an effort to make the contents searchable. The project, Google Book Search, has come under fire from some authors who are unhappy that Google is including books still protected by copyright without expressly gaining permission from the authors. Using the new shape-matching technology to make hand written manuscripts searchable is unlikely to meet with similar criticism, since the documents are historical and wouldn’t be protected by copyright.


IDG UK Sites

Best Black Friday 2014 tech deals: Get bargains on smartphones, tablets, laptops and more

IDG UK Sites

What the Internet of Things will look like in 2015: homes will get smarter, people might get fitter

IDG UK Sites

See how Trunk's animated ad helped Ade Edmondson plug The Car Buying Service

IDG UK Sites

Yosemite tips: Complete Guide to OS X Yosemite