acm-header
Sign In

Communications of the ACM

ACM TechNews

AI Can Help Historians Restore Ancient Texts


View as: Print Mobile App Share:
The Celsus Library in the ancient city of Ephesus, Turkey

"We hope that the way we've designed it, it's going to be easy for [a] historian to use." -Yannis Assael, DeepMind

Credit: Mazur Travel/Shutterstock

Ithaca, a new DeepMind algorithm dedicated to helping historians restore ancient Greek inscriptions, is yielding restorations with accuracies of up to 72%. Researchers at the U.K.-based artificial intelligence company trained Ithaca on roughly 60,000 ancient Greek texts, then concealed certain characters and compared Ithaca's predictions for the "missing" text with actual inscriptions. They then used some 8,000 inscriptions to test Ithaca's performance alone, or in conjunction with two historians, which respectively yielded 62% and 72% accurate restorations.

Ithaca also could predict where in the Mediterranean a text was written 71% of the time, and could date the texts to within three decades of their true age.

From New Scientist
View Full Article

 
Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account