Biblissima+ (Observatoire des cultures écrites, de l’argile à l’imprimé), supported by the Campus Condorcet, is one of the EquipEx+ structuring facilities for research funded under the Investissements d'avenir program. It takes over from ÉquipEx Biblissima (Bibliotheca bibliothecarum novissima) and creates a multi-polar digital infrastructure for fundamental research and services dedicated to the history of the transmission of ancient texts, from the first Mesopotamian clay tablets to the first printed books, on all media and in all languages.

Biblissima+ is structured around 7 areas of expertise or « clusters » organized according to the data life cycle. Cluster 3 “Artificial Intelligence for pattern and handwriting recognition”, coordinated by Dominique Stutzmann (directeur de recherche, IRHT - CNRS) and Daniel Stökl Ben Ezra (directeur d'études, EPHE-PSL, AOROC UMR 8546), focuses on the use of artificial intelligence for the recognition of forms, characters, and handwriting, enabling the study of ancient handwritten and printed books, sigillography, numismatics, and heraldry. Biblissima+ will be dealing with massive and diverse data, in terms of materials, languages, and writing systems. The cluster's work will be carried out in close collaboration with other clusters to formalize text, exploit images, read and understand text. The cluster aims to open up access to infrastructures and data and to provide users with the most powerful and up-to-date tools possible while respecting the finesse and diversity of historical research.

Biblissima+ is involved in the discussion about the future of training offered to communities focused on the analysis of medieval manuscripts using artificial intelligence. In this context, this Cluster 3 day is organized as a seminar reflecting on the future of doctoral training at the European level. A seminar with key members of this community will take place throughout the day around the topic "a doctoral network for the future : M3D – Medieval Manuscripts – From Material to Digital and Back," concluding with a public roundtable at the end of the day. 

 Biblissima+ is supported by a grant of Agence Nationale de la Recherche (Programme d’investissements d’avenir / France 2030, ANR-21-ESRE-0005).

The European Middle Ages are often misrepresented or misused, both in scholarship and in society/media. The increasing availability of digital data and images of Medieval Manuscripts, our main source for information about the Middle Ages, presents an opportunity to fundamentally change this narrative and exponentially expand our knowledge. However, there is a challenge: scholarship lacks the know-how, both technical and conceptual, to deal with this mass of new data and fully realize its potential. Both creation of and research on these data happens in a fragmented way, without clear direction or conceptual and methodological standardization. To overcome this problem, we need to forge stable and sustainable interdisciplinary bonds, create a new paradigm for the use of medieval manuscripts as historical sources and train a new generation of scholars to use the data, apply the technologies, and ask new questions. A doctoral network is the perfect environment to achieve this. With this network, we can achieve a new paradigm for the study and representation of the Middle Ages, centered on medieval manuscripts, our richest and currently most underexplored source of data for the period.

Medieval Manuscripts, i.e. the written cultural heritage of the European Middle Ages, are a dynamic, interdisciplinary, and highly symbolic field of research in the European and global context, in which Artificial intelligence (AI) creates a revolution, not only for scholars, but above all in how people see, interact, interpret, reproduce, spread, and take possession of the past and its representations. The abundance of forgeries and fakes, supporting conspiracy theories, based on real or invented historical events and objects, and erasing the historical complexity, is a major challenge. With the increased digitization of Culture Heritage collections since the 1990s and even further now with the European Digital Decade, the availability of source material both for “human consumption” and to train AI image generators is immense and increases the risks of misrepresentation and misuse of the European Middle Ages. Whether on purpose or through an overreliance on tools, possible errors and the risk of misinterpretations of texts conveyed through computer vision, image analysis, handwritten text recognition (HTR), automated translation, or hallucinated references, create, at worse, new errors and misleading information and, at best, a flow of repeated obsolete interpretations due to the inherent biases in models with legacy data. The Western Middle Ages are in this regard specific in the way the period is used in the public discourse. Its exploitations, including political ones, have therefore a significant impact, serving as an idealized past as the source of heroic or national legends (Blanc and Naudin 2015; Cooper 2023) or even as a model a society thought of as “white” and homogeneous (“pure”) with strong religious authority under the supervision of the (Christian) Church on the one hand, and as a negative portrayal – the words “medieval”, “dark”, “barbaric”, representing an oppressive, immutable society, as a rhetoric repellent on the other. Numerous allusions and visual references abound, as the Middle Ages are a recurring element in both popular and scholarly culture (e.g., Il nome della rosa, The Lord of the Rings, Harry Potter, Games of Thrones, and the opera Written on Skin by G. Benjamin). The association of Europe with its Middle Ages is also exploited beyond the Western context (e.g., Crusades).
This great societal challenge requires a new generation to achieve an enhanced level of expertise jointly in several domains: on (a) MSS, or manuscript studies, i.e. historical sources and memorial stakes; (b) CS, or computer science, esp. artificial intelligence (AI), large language models (LLMs), image analysis and computer vision; (c) DHist, or digital history, i.e. information modeling and the specific requirements in the analysis of digital data representing historical artifacts. Yet no doctoral school in Europe can achieve this level of expertise and interdisciplinary education.

How to create a community of early career and advanced researchers, capable of addressing simultaneously the procedures, requirements and evolution of Manuscript Studies, Digital History, and Computer Science? Three categories of research objectives contribute to attaining this ORG: disciplinary, interdisciplinary, and societal.
But first: Why medieval manuscripts? They are the main and richest witness to medieval culture and mindset, and an inexhaustible source of information about the past. Various disciplines concur to their understanding: philology, diplomatics, paleography, codicology, history, art history, literary and cultural studies, which, however, form separate fields (Rouse-Rouse-Baswell 2011; Mostert 2012; Bausi et al. 2015). Indeed, medieval manuscripts are multifaceted objects. Long seen separately by historians, philologists and art historians only as either text or image carriers, they are recognized in their textual, physical and historical complexity. Texts are now analyzed in their fluidity and through the processes of their transmission (“textual tradition”), which entails copying, rewriting, interpolating in a situation where each text carrier may transmit a very popular work, but is still, each time, a very unique witness. The physical complexity has first given rise for a more holistic approach of text and image relations within a volume. Now it is the object of dedicated discipline, named “codicology” or “archeology of the book”, and is part of a renewed approach in manuscript studies and diplomatics, where the physical aspects are considered constitutive elements of the message (Barret, Stutzmann and Vogeler, 2016; Treharne 2021) and all features of a manuscript work together to communicate ideas and transmit culture (Andrist-Canart-Maniaci 2013; Quenzer 2021). Their physicality is integrated in social and economic history and global approaches, for instance in the study of oversea trade of pigments and materials from Africa () or import of techniques from Asia such as paper production (), as well as in environmental and ecological perspective, e.g., meat consumption and parchment production and understanding intellectual production and transmission with biological models (Camps-Randon-Furling 2022; Kestemont et al. 2022). The contextual information and long history (origin, provenance) and successive transformations and evolutions (damage, deletions and additions, repurposing) of medieval manuscripts, and their actors are more fully analyzed in a dynamic perspective. Each combination and presentation of texts in a manuscript, whether a carefully designed luxury copy or a random personal notebook, influences the texts in question, their integrity, shape, interpretation, and reception (Friedrich-Schwarke 2016; Nichols-Wenzel 1996).

Yet, the very dynamics and reciprocal interactions at all stages of the manuscript history between texts and their presentations are not formalized and cannot be operationalized. We now realize that the projection of the scholars’ traditional categories onto the manuscript complexity obfuscates rather than elucidates, and impose a strong bias in the available data on manuscripts and skew our interpretative framework, making the larger question of the mechanics of medieval knowledge transmission currently unanswerable (Cermanová-Zurek 2021; Pratt et al. 2017).

Embedded in a trend that applies transdisciplinary digital methods to large corpora of manuscripts to increase and diversify their value as historical sources (Maniaci 2022; Stokes 2021), M3D wants to address medieval manuscripts in their complexity and first identifies three crucial domains which are a source for uncertainty and confusions and for which researchers are not sufficiently equipped, in theory, in methods, in scholarship, and in processing techniques:
(a) Visual complexity and information organization in medieval manuscripts;
(b) Textual interconnections between texts, works and text carriers;
(c) Identification of people and places of the Middle Ages.

In all three, hitherto insufficient scholarly knowledge, lack of understanding and capacity of modeling, and limitation of tools are a limitation for advancing our understanding and transmitting a deeper understanding of the past. They are therefore the focus of three Work Packages.

The conclusive round table discusses "Medieval Manuscripts From Material to Digital: the Future of Doctoral Training in Europe". Attendance is free, but registration is required (follow this link). The round table will be held in English.

Along with disciplinary contents, the participants will discuss organizing a doctoral training promoting transferable and intersectorial skills and inclusivity with a top-quality supervision.

Announced participants are:

Shari Boodts  Professor at the Radboud University in Nijmegen.

Lucie Doležalová  Professor at Univerzita Karlova (Charles University) in Prague.

 Mark Faulkner  Professor at Trinity College in Dublin.

 Franz Fischer  Professor at the Università Ca' Foscari in Venice.

 Alicia Fornés  Senior researcher à l'Universitat Autònoma de Barcelone (UAB).

 Christopher Kermorvant  CEO of TEKLIA.

 Mike Kestemont  Professor at the university of Antwerp.

 Peter Stokes Professor at EPHE-PSL.

 Dominique Stutzmann  Director of research at IRHT - CNRS.

 Georg Vogeler  Professor at the Karl-Franzens-Universität in Graz.