INTORNO A L’ ‘EPISTOLA A CANGRANDE’ AL VAGLIO DELLA ‘COMPUTATIONAL AUTHORSHIP VERIFICATION’: IMPOSTAZIONI E PROBLEMATICHE

The paternity of the ‘Epistola a Cangrande’ is probably the most salient open problem in Dante’s
studies, such that any proposed solution must be thoroughly examined. In this work, the articles from
Corbara et alii, ‘L’ ‘Epistola a Cangrande’ al vaglio della ‘Computational Authorship Verification’’
and ‘The ‘Epistle to Cangrande’ Through the Lens of Computational Authorship Verification’, are
analyzed and commented using arguments from the theoretical framework of Machine Learning
(ML) and Authorship Verification (AV) and its literature. The reliability of any AV model is dependent
on its training texts and the features chosen to characterize them, and by presenting limitations of
the used models, which showcase some inherent subjectivities of ML, we conclude that the problem of
the Epistola’s paternity remains thus open.