Greek Medieval and Modern Minuscule

This model is intended for medieval and modern Greek Minuscule manuscripts. The initial data set consisted of ten manuscripts of dates ranging from the 10th to 19th centuries. Several of these manuscripts were the product of more than one scribe, however. On which account, this model is also trained on at least 15 unique hands. Their texts were all from among the Old and New Testament or the Testaments of the Twelve Patriarchs. Spaces have been included between words in the training data even when not present in the manuscript (as was the case in Cambridge University Library manuscript Ff 1.24). Thus, the model attempts to separate words even for scripta continua manuscripts. However, this model has been trained to extract the text only and no other features. Diacriticals, accents, punctuation etc. have been excluded. Capitalization has likewise been ignored. The model has not been trained to resolve abbreviations, alphabetic representation of numerals or nomina sacra. It has, however, been trained to resolve ligatures. The list of manuscripts on which this model was trained is as follows: Bodleian Library - Barocci 133 Bodleian Library - Holmes 94 Bodleian Library - Holmes 155 Bodleian Library - Smith 117 British Library - Harley 7522A Cambridge University Library - Ff 1.24 Cambridge University Library - Oo.VI.91,8 Queens College, Oxford - 214 Trinity College, Cambridge - O.4.24 Trinity College, Cambridge - B.10.3 *Note: transcriptions of the entire codices were not always used.