All Fire, all Flames (AFAF): from Philology to artificial Intelligence and Back. Recovering the MS L.II.14 of the Biblioteca Nazionale Universitaria di Torino damaged by the Fire of 1904

Progetto di Ricerca di Interesse Nazionale

Principal Investigator: Marco Maulu (Università degli Studi di Sassari)

The "All fire, all flames" project (AFAF) is aimed at stemming from the two-year project entitled "RESCAPÉ: L’approche numérique face au défi du feu", funded by Biblissima+ (CNRS, France) 2021-22, PI Marco Maulu. RESCAPE exploits the artificial intelligence applied to automatic handwriting recognition (Handwritten Text Recognition = HTR) and specifically to the MS Torino, BNU, L.II.14 (= T, around 1311), an extremely important French codex damaged by the fire of 1904. In RESCAPE a complete automatic and semi diplomatic transcription of the MS in XML-TEI format is obtained by using Escriptorium, an innovative tool developed at the Ecole Pratique des Hautes Etudes (an example of T segmentation obtained via Escriptorium can be seen here:). For the results of the text prediction obtained 20/11/23, 18:59 PRIN 2022 PNRRPRIN 2022 PNRR https://prin2020.cineca.it/simbad/php5/sito/vis_modello.php?info= with standard HTR models, see ). AFAF starts from the result of the HTR obtained thanks to RESCAPE: AFAF is therefore going to exploit the semi diplomatic transcription of T in .xml-alto and already corrected (fine-tuned) by the human being. This file will then be ready for a paper, interpretative edition of the whole MS, which is the main objective of AFAF. A descriptive catalogue of the beautiful 86 miniatures of T is also expected. The skills in X-Ray Imaging of Catania ISPS-CNR AFAF UR will improve the readability of the most damaged portions (notably vols. 1 and 4).

Due to the high cohesion of its texts, T can be considered to all intents and purposes as an anthological codex, a fact that justifies itself the need of editing all the volume, without neglecting the absence of digital reproductions and the tendency to overlook this witness due to its state of conservation.

Currently, digital and traditional editions have little dialogue: many digital editions in .xml obtained through HTR or OCR (Optical Character Recognition) are stored in "hosting repositories" such as GitHub. This can cause maintenance and perpetuation problems of the data and no knowledge of such editions by a non-DH educated public. Furthermore, the results obtained through the HTR are often not improved due to lack of funds, interest, adequate editorial and philological skills, etc. AFAF intends to put computational and traditional philology in communication, creating a ground-breaking synergy between these two complementary skills to obtain an excellent result in faster times than those that a traditional edition would allow on a vast corpus such as T. Such innovative “hybrid” methodology could open new perspectives as a template for textual criticism applied to corpus hitherto difficult to approach by traditional methodologies. The contribution of two new research contracts is foreseen. We look forward to plan two conferences in Naples and Sassari as an opportunity for methodological reflection on Digital Humanities applied to the edition of medieval texts.

Unità di Ricerca

• Università degli Studi di Sassari: Marco Maulu (Principal Investigator)

• Università degli Studi di Napoli Federico II: Paolo Di Luca

• Università degli Studi di Catania: Stefano Rapisarda

• Università degli Studi della Basilicata: Carlo Alberto Beretta