PYTHIA: a deep learning model for the automatic restoration of Greek inscriptions

Pythia is the first ancient text restoration model that recovers missing characters from a damaged text input using deep neural networks. It offers an automated aid to the epigraphic restoration of fragmentary inscriptions, providing historians with multiple textual restorations and the confidence level for each hypothesis.
Pythia was trained on a dataset of machine processed ancient Greek inscriptions. Pythia's architecture works at both the character- and word-level, effectively handling long-term context information, and dealing efficiently with incomplete word representations. This makes it applicable to all disciplines dealing with fragmentary texts, and to any language, ancient or modern.



University of Oxford

Project Team

Thea Sommerschield, University of Oxford
Yannis Assael, DeepMind
Jonathan Prag, University of Oxford