‘Pythia’ helps find missing parts of Ancient Greek inscriptions

‘Pythia’ helps find missing parts of Ancient Greek inscriptions

Researchers at Oxford University and DeepMind Technologies, a UK-based Artificial Intelligence (AI) company owned by Alphabet Inc. – a conglomerate best known as the parent company of Google – have created “Pythia,” an ancient text restoration system that outperforms experts in guessing the missing text from partially destroyed Ancient Greek inscriptions.

The fully automated system, the first of its kind, fills the missing characters, or even whole words, by making alternative proposals. It can thus become a very useful tool for experts trying to read and restore ancient inscriptions, dozens of which are discovered each year to be added to the existing trove of several thousands.

The passage of time has not been kind to many of those inscriptions, which have either been partially destroyed or the material they have been written in corroded. Deciphering them is not easy and requires patience, knowledge, comparison with other texts and a good deal of intuition.

The researchers, a Greek, Yannis Assael, among them, trained Pythia to predict missing letters of words using deep neural networks. (Neural networks are sets of algorithms, patterned after the human brain, that are trained to recognize patterns). Pythia learned to recognize such patterns on some 35,000 inscriptions ranging from the 7th centuty BC to the 5th century AD (1,500 to 2,700 years old) that contain over 3 million words.

When shown an incomplete inscription, Pythia makes up to 20 hypotheses, that is, proposes up to 20 different letters or words, leaving it to the experts to make the final choice.

In a test of the system, both Pythia and inscription experts, or epigraphists (actually, PhD students at Oxford), were set the task of “filling in the blanks” in 2,949 inscriptions. Pythia had a 30.1 percent character error rate, compared to the human experts’ 57.3 percent. Moreover, while the experts needed two hours to decipher 50 inscriptions, Pythia needed a few seconds. The result was a clear demonstration of the potential of AI in digital epigraphy, said Thea Sommerschield of Oxford, one of three co-authors of a paper on Pythia (“Restoring ancient text using deep learning: a case study on Greek epigraphy”) Philippa Steele, a Professor of Classics at Cambridge, agreed that “Pythia” can contribute significantly to the restoration of Ancient Greek inscriptions, although its pattern-detecting power will always need to be supplemented by the insights of experts. Pythia could also be used in related fields, such as papyrology and codes, she said.

Yannis Assael, one of the three authors of the paper, graduated with an Applied Informatics Degree from the University of Macedonia, in Thessaloniki, in 2013, did graduate studies at University of London’s Imperial college and is currently pursuing his Ph.D. in Computer Science at Oxford, specializing in Machine Learning. Since 2015, he is also an AI researcher at DeepMind.

Demis Hassabis, son of a Greek Cypriot father and a Singaporean Chinese mother, and well-known as a video game developer and a world-class gamer himself, was one of the three co-founders of DeepMind Technologies in 2010, alongside Shane Legg and Mustafa Suleyman, and is the company’s CEO. DeepMind became a Google subsidiary in 2014 and has research centers in Canada, France and the United States.


Subscribe to our Newsletters

Enter your information below to receive our weekly newsletters with the latest insights, opinion pieces and current events straight to your inbox.

By signing up you are agreeing to our Terms of Service and Privacy Policy.