Computational Transcription of Medieval Hebrew Manuscripts and Crowdsourcing their Corrections

We will present initial results on two computational projects on Medieval Hebrew manuscripts. The first, Sofer Mahir, applies an HTR (handwritten text recognition) pipeline constructed at Scripta-PSL to the major manuscripts of the classical compositions of the tannaitic period of Rabbinic Judaism. In the frame of the second project, Tikkoun Sofrim, which applies the pipeline to manuscripts of early Medieval Tanhuma-Yelamdenu Midrashim, we have developed a crowdsourcing platform that permits citizen scientists to suggest corrections to the automatic transcription.