Tel Aviv, PTI
A significant development has emerged for historical researchers as students at Ben-Gurion University of the Negev have utilized artificial intelligence to restore illegible letters and words in ancient Hebrew and Aramaic inscriptions. Every year, archaeologists unearth numerous ancient texts written in Hebrew and Aramaic across the Near East, which are crucial for understanding the region’s cultural and historical heritage. However, damage over time has rendered many of these texts indecipherable, posing challenges for scholars.
BGU’s innovative approach holds promise for revolutionizing the field of epigraphy, which involves identifying, classifying, and interpreting inscriptions on ancient artifacts like coins, monuments, and scrolls. Professor Mark Last, who supervised the students’ project, believes this breakthrough has the potential to transform epigraphy. Traditionally, reconstructing missing parts of damaged inscriptions relied on manual methods prone to errors.
The students from the university’s Department of Software and Information Systems Engineering approached the challenge as an extended masked language modeling task, a type of natural language processing task that builds upon the concept of masked language modeling. This technique, commonly used in pre-training large-scale language models, enables more accurate reconstruction of ancient texts and holds promise for adaptation to other morphologically rich ancient languages. The project not only aids historians in reconstructing ancient texts more accurately but also offers potential applications in diverse linguistic contexts.