International Journal of Advanced Multidisciplinary Application

Restoring Fragmented Heritage: Deep Learning Applied to Ancient Epigraphy

Author(s):Arjun Mehra, Prof Yasmine Al-Rashid

Affiliation: Centre for Digital Humanities, University of Hyderabad, Hyderabad, India

Page No: 53-58

Volume issue & Publishing Year: Volume 3, Issue 7, July 2026

published on: 2026/05/07

Journal: International Journal of Advanced Multidisciplinary Application.(IJAMA)

ISSN NO: 3048-9350

DOI:

Download PDF Cite this article

Abstract:
The degradation of ancient inscribed surfaces — through mechanical fracture, chemical weathering, and deliberate obliteration — has rendered substantial portions of the world's epigraphic record unreadable by conventional philological means. This project bridges computational linguistics, machine learning, and heritage science to reconstruct damaged ancient inscriptions using deep learning architectures trained on attested linguistic corpora. By encoding known phonological, morphosyntactic, and graphemic constraints derived from corpus linguistics into transformer-based and recurrent neural network models, the system generates probabilistic predictions for lacunae — gaps, breaks, and abraded zones — on weathered stone, metal, and ceramic surfaces. Validation against three expert epigraphists' reconstructions across a 1,200-inscription test set demonstrates character-level accuracy of 93.6% for Latin and Greek scripts and 81.4% for underdeciphered scripts including Linear B and the Indus Valley script. This approach accelerates historical translation workflows by an estimated 4.7-fold and provides a systematic framework for preserving vulnerable cultural heritage data in machine-readable, open-access formats.

Keywords: ancient epigraphy, deep learning, transformer model, BiLSTM, lacuna reconstruction, heritage preservation, natural language processing, corpus linguistics, photogrammetry, open-access platform

Reference:

• [1] Assael, Y., Sommerschield, T., Shillingford, B., Bordbar, M., Pavlopoulos, J., Chatzipanagiotou, M., ... & de Freitas, N. (2022). Restoring and attributing ancient texts using deep neural networks. Nature, 603(7900), 280–283.
• [2] Biagetti, E., Sommerauer, P., & Kuhn, J. (2020). Automatic detection of lacunae in ancient inscriptions. Journal of Data Mining and Digital Humanities, 2020, 1–18.
• [3] Chadwick, J., & Ventris, M. (1973). Documents in Mycenaean Greek (2nd ed.). Cambridge University Press.
• [4] Corpus Inscriptionum Latinarum (CIL). (2023). CIL Digital Edition v3.1. Berlin-Brandenburgische Akademie der Wissenschaften. https://cil.bbaw.de
• [5] Dieleman, W. (2023). Multispectral imaging of Ptolemaic temple inscriptions: Recovery of obliterated divine epithets at Dendera. Journal of Egyptian Archaeology, 109(1), 77–104.
• [6] Farmer, S., Sproat, R., & Witzel, M. (2004). The collapse of the Indus-script thesis: The myth of a literate Harappan civilization. Electronic Journal of Vedic Studies, 11(2), 19–57.
• [7] Gnanadesikan, A. E. (2009). The Writing Revolution: Cuneiform to the Internet. Wiley-Blackwell.
• [8] Inomata, T., & Houston, S. (Eds.). (2001). Royal Courts of the Ancient Maya (Vol. 1). Westview Press.
• [9] Kamp, A., Peake, N., & Schulz, A. (2012). Probabilistic sequence models for epigraphic reconstruction. Proceedings of LREC 2012, Istanbul, 3412–3417.
• [10] Mahadevan, I. (1977). The Indus Script: Texts, Concordance and Tables. Archaeological Survey of India.
• [11] Packard, D. W. (1968). Computer-assisted morphological analysis of ancient Greek. Proceedings of the International Conference on Computational Linguistics (COLING), 1–16.
• [12] Parpola, A. (1994). Deciphering the Indus Script. Cambridge University Press.
• [13] Sommerschield, T., Assael, Y., Pavlopoulos, J., Stefanak, V., Senior, A., Dyer, C., ... & Blunsom, P. (2023). Machine learning for ancient languages: A survey. Computational Linguistics, 49(3), 703–751.
• [14] UNESCO. (2021). Endangered World Heritage Sites: State of Conservation Report 2021. UNESCO World Heritage Centre.
• [15] Woodard, R. D. (Ed.). (2004). The Cambridge Encyclopedia of the World's Ancient Languages. Cambridge University Press.