In an unprecedented development for students, historical researchers at Ben-Gurion University of the Negev, Tel Aviv [Israel], have used the power of artificial intelligence to restore unreadable letters and words in ancient Hebrew and Aramaic inscriptions. Every year, archaeologists unearth ancient objects. Texts written in Hebrew and Aramaic in the Near East. These inscriptions are invaluable for understanding the rich cultural and historical heritage of the region, however, many of these texts have suffered damage over time, making them difficult for scholars to decipher. Natural disasters, political conflicts and the ravages of time have taken their toll on these ancient artefacts, but BGU's innovative approach can revolutionize the field of epigraphy by identifying, classifying inscriptions found in ancient artefacts like coins, monuments, etc. And there is a science of interpretation."This breakthrough has the potential to revolutionize the field of epigraphy," said Professor Mark Last, who supervised the students' project. "Not only can we reconstruct ancient texts. can help historians more accurately, but I also believe that this model can be adapted to other morphologically rich ancient languages. Traditionally, archaeologists relied on time-consuming manual methods to reconstruct missing parts of damaged inscriptions. However, those methods are prone to errors. The students from the university's Department of Software and Information Systems Engineering who took up the project approached the challenge as "an extension of the masked language modeling task." It refers to a specific type of Natura language processing task that is based on the concept of masked language modeling, a technique commonly used in pre-training large-scale language models.Damaged content may include single characters, character n-grams (partial words), single complete words, and multi-word n-grams. Under the leadership of Last, graduate students Niv Fono, Harel Moshayoff, Eldar Karol, and Itay Assarf implemented a mask language modeling. The approach to corrupt inscriptions in Hebrew and Aramaic involved training the system on a dataset containing 22,144 sentences of the Old Testament and testing it on an additional 536 sentences, achieving notable success, by employing a suite of word and character prediction models. , they were able to achieve high accuracy in restoring damaged text. Their model, called "Ambiable", was presented to the European Chapter at the meeting of the Association for Computational Linguistics in March. "We can help those historians Who have dedicated their lives to accurately recreating these ancient texts." As much as possible," Last said, "Furthermore, I believe this mode can be extended to cover other morphologically rich ancient languages ​​(ANI/TPS).