Wordspotting
I still can't glean exactly how this would work, but it sounds like an important breakthrough for people who spend a lot of time searching for information in handwritten manuscripts. This would include not just scholars but also the millions of amateur genealogists out there (though it's unclear whether this will ever be anything more than an expensive tool for academic specialists). I do like the term "wordspotting" though.
... But what about documents that mice like to munch -- in attics, basements, and stuffy storage rooms -- handwritten documents from the past, such as Einstein's original Theory of Relativity; the Bill of Rights; or the Emancipation Proclamation?
Keeping those treasures intact is crucial to historians and researchers who need to search handwritten documents -- such as 140,000 pages of George Washington's personal papers in the Library of Congress -- for pieces of the past.
University of Massachusetts Amherst (UMass) information technology researchers say they've developed a first-of-its kind manuscript-searching system to aid that effort.
... Using 1,000 scanned pages of Washington's manuscripts, Manmatha and computer science graduate students Toni Rath and Victor Lavrenko created a demonstration of their search tool. Instead of Trainspotting, the software program does wordspotting. Type a term, such as "Washington" or "Virginia," and it produces a list of ranked pages showing where the term appears.
"The basic idea is analogous to searching text documents in one language -- say French -- using queries in another language -- say English," Toni Rath explained. "This is usually done by learning models from documents written in both languages."
[Tags: Einstein relativity Bill of Rights constitution manuscripts research history George Washington documents technology search engines genealogy wordspotting Massachusetts Amherst University of Massachusetts Emancipation Proclamation Lincoln Virginia]
... But what about documents that mice like to munch -- in attics, basements, and stuffy storage rooms -- handwritten documents from the past, such as Einstein's original Theory of Relativity; the Bill of Rights; or the Emancipation Proclamation?
Keeping those treasures intact is crucial to historians and researchers who need to search handwritten documents -- such as 140,000 pages of George Washington's personal papers in the Library of Congress -- for pieces of the past.
University of Massachusetts Amherst (UMass) information technology researchers say they've developed a first-of-its kind manuscript-searching system to aid that effort.
... Using 1,000 scanned pages of Washington's manuscripts, Manmatha and computer science graduate students Toni Rath and Victor Lavrenko created a demonstration of their search tool. Instead of Trainspotting, the software program does wordspotting. Type a term, such as "Washington" or "Virginia," and it produces a list of ranked pages showing where the term appears.
"The basic idea is analogous to searching text documents in one language -- say French -- using queries in another language -- say English," Toni Rath explained. "This is usually done by learning models from documents written in both languages."
[Tags: Einstein relativity Bill of Rights constitution manuscripts research history George Washington documents technology search engines genealogy wordspotting Massachusetts Amherst University of Massachusetts Emancipation Proclamation Lincoln Virginia]
0 Comments:
Post a Comment
Links to this post:
Create a Link
<< Home