English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Good OCR results for historical printings rely on the availability ofrecognition models trained on diplomatic transcriptions as ground truth, whichis both a scarce resource and time-consuming to generate. Instead of having totrain a separate model for each historical typeface, we propose a strategy tostart from models trained on a combined set of available transcriptions in avariety of fonts. These \emph{mixed models} result in character accuracy ratesover 90\% on a test set of printings from the same period of time, but withoutany representation in the training data, demonstrating the possibility toovercome the typography barrier by generalizing from a few typefaces to alarger set of (similar) fonts in use over a period of time. The output of thesemixed models is then used as a baseline to be further improved by both fullyautomatic methods and semi-automatic methods involving a minimal amount ofmanual transcriptions. In order to evaluate the recognition quality of eachmodel in a series of models generated during the training process in theabsence of any ground truth, we introduce two readily observable quantitiesthat correlate well with true accuracy. These quantities are \emph{meancharacter confidence C} (as given by the OCR engine OCRopus) and \emph{meantoken lexicality L} (a distance measure of OCR tokens from modern wordformstaking historical spelling patterns into account, which can be calculated forany OCR engine). Whereas the fully automatic method is able to improve upon theresult of a mixed model by only 1-2 percentage points, already 100-200hand-corrected lines lead to much better OCR results with character error ratesof only a few percent. This procedure minimizes the amount of ground truthproduction and does not depend on the previous construction of a specifictypographic model.

Automatic quality evaluation and (semi-) automatic improvement of OCR models for historical printings