Google is funding research aimed at making handwritten documents searchable. The Dublin City University project is a rare external collaboration for Google, which in the past has preferred to simply buy in the expertise it needs.
Computer scientists are adapting technology originally developed to recognise objects like cars and people in video. Team leader Professor Alan Smeaton told The Register: “We stumbled upon the idea of using the algorithms for handwriting.”
When the system is given an example of a word in someone's handwriting, it can then search through documents written by that person and find other instances, adapting to variation in style. The approach has already been successfully tested on George Washington's personal diaries – every appearance of the word "battle" can be quickly accessed, for example.
The researchers say before now this kind of material has only been accessible in digital libraries one page at a time, which is slow and cumbersome, or is kept behind closed doors.
Google has provided enough funding for the team at Dublin, and its partners at the Universities of Buffalo and Massachusetts to work on the problem for at least a year. It's hoped the tools the group develop will become a key part of Google's stated aim of digitising the world's libraries.
Professor Smeaton said: “This will make historical manuscripts searchable for scholars and others in a way that has never been possible before.”®