The data_train contains sentences taken from 50 Italian novels published in XIX and early XX centuries. The data are organized according to the SpaCy requirements. The NER labels have been attributed according to an extensive set of rules to identify characteres, places, dates ... This dataset has been used to train the model ItLit800 whose aim is to extend the capabilities of the standard SpaCy NER pipeline for the Italian language.