The Tel Aviv University team of Professor Nachum Dershowitz creates and utilizes advanced computational techniques to find solutions for complex problems of digital Hebrew paleography and for analyzing medieval manuscripts in Hebrew characters. Deep machine learning is used to build state-of-the-art algorithms for automatic dating of undated manuscripts, recognition of the region of copying, and automatic recognition of script types and modes. The research involves synergetic feedback between human paleographers and computer scientists to create new tools for Hebrew paleography, including algorithms for clustering and sub-clustering medieval manuscripts within known script type-modes, extraction of paleographical features, and new tools for page segmentation and layout recognition. The machine classifications are translated into human-understandable terms to ensure meaningful feedback.
