New Developments in Kraken and eScriptorium

Par Benjamin Kiessling (PSL) et Peter Stokes (EPHE).

Recent releases of Kraken (v4) and eScriptorium introduce a number of new features that improve user experience and performance. The presentation will introduce the most important ones such as the new training library, binary datasets, and new layer types for Kraken, and annotation and text search for eScriptorium, as well as integration of both into Biblissima+. We will elaborate how these impact the use of the software in a variety of contexts, such as institutional and individual use, differences in dataset and target corpus size, etc. In addition, we will look briefly at subsystems in development such as a new algorithm for trainable reading order.