Papers › Restoring and attributing ancient texts using deep neural networks

Restoring and attributing ancient texts using deep neural networks

9 Mar 2022Nature 2022 3archive 2025-07-28

Yannis Assael, Thea Sommerschield, Brendan Shillingford, Mahyar Bordbar, John Pavlopoulos, Marita Chatzipanagiotou, Ion Androutsopoulos, Jonathan Prag, Nando de Freitas

Ancient history relies on disciplines such as epigraphy—the study of inscribed texts known as inscriptions—for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian’s workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25% to 72%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.

PaperPDFCode

Code

deepmind/ithaca officialmentioned in paperjax report
sommerschield/iphi officialmentioned in paper report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Ancient Text RestorationAttribute

Datasets

Introduced by this paper, per the archive.

I.PHI

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Ancient Text Restoration I.PHI Ancient historian and Ithaca CER (%) 18.3 #1 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ancient historian and Ithaca Top 1 (%) 71.7 #1 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ancient historian and Ithaca Top 20 (%) 78.3 #1 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ithaca CER (%) 26.3 #2 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ithaca Date (Years) 29.3 #2 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ithaca Region (Top 1 (%)) 70.8 #2 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ithaca Region (Top 3 (%)) 82.1 #2 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ithaca Top 1 (%) 61.8 #2 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Pythia CER (%) 47.0 #3 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Pythia Top 1 (%) 32.6 #3 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Pythia Top 20 (%) 53.9 #3 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ancient historian CER (%) 59.6 #4 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Ancient historian Top 1 (%) 25.3 #4 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Onomastics Date (Years) 144.4 #5 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Onomastics Region (Top 1 (%)) 21.2 #5 of 5 Archive leaderboard report
Ancient Text Restoration I.PHI Onomastics Region (Top 3 (%)) 26.5 #5 of 5 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections