Artificial intelligence software can help historians interpret and date ancient texts by reconstructing works that have been destroyed over time, according to research published in Nature.
A team of computer scientists and experts in classical studies led by DeepMind and Ca 'Foscari University in Venice trained a neural network to be able to restore inscriptions written in ancient Greek between the 7th century BC. and 5th century AD
The model, named "Ithaca", can also estimate when the text was written and where it might come from. So researchers can retrieve texts found in broken pottery pieces, or blurred texts found in various inscriptions.
First, the text should be transcribed by scanning an image of an old object. It is then fed to Ithaca for analysis. Works by predicting lost or blurred characters and restores words. The software creates and ranks a list of its top predictions. Archaeologists can then read them and judge whether the model's predictions are accurate or not.
Of course, better results are achieved with the cooperation of man and machine. When archaeologists worked alone, they were 25 percent accurate in combining ancient texts. In partnership with Ithaca the level of accuracy rose to 72 percent. The performance of the machine learning model alone is about 62 percent accurate. It can also pinpoint 71 percent of the location where the text was written and can date the works 30 years from their creation between 800 BC. and 800 AD.
Ithaca trained on more than 63.000 Greek inscriptions containing more than three million words from the Packard Humanities Institute's Searchable Greek Inscriptions repositories. The team filled in parts of the text and instructed the model to fill in the blanks.
Google's DeepMind is now adapting its model to other types of ancient writing systems, such as Akkadian developed in Mesopotamia, the language of ancient Egypt, the Mayan language of Central America, and ancient Hebrew.
Registration in iGuRu.gr via email
Follow us on Google News