Tuesday, June 11, 12:30-1:15pm, Class of 1978 Orrery Pavilion
Artificial Intelligence, Handwritten Text Recognition and Access to Medieval Sources
Artificial Intelligence has unlocked the access to medieval sources! In several research projects, Handwritten Text Recognition (HTR) and Image classification, mainly based on Machine learning provide new ways of accessing, analyzing and understanding the past.
This talk will present (1) the results of the European funded HIMANIS research project, finished in 2018, in which the partners were able to achieve the text indexing of more than 80’000 pages and advance writer identification, (2) the ongoing work in the research project HORAE, based on IIIF access to images, image classification (miniature, text, calendar, etc.), text recognition, and text reuse identification, and (3) the rationale of HOME, aiming at producing a meaningful research environment to give access to charters and cartularies in Europe (HTR, semantic information retrieval, and named entity recognition and linking). Open to everyone, this presentation is particularly connected to the courses “Digital Surrogates: Representation, Engagement, and Meaning”, “Linked Data for the Humanities”, and “Text Analysis”.
Dominique Stutzmann is a senior researcher at IRHT-CNRS, member of Comité International de Paléographie Latine.
Wednesday, June 12, 12:30-1:15pm, Class of 1978 Orrery Pavilion
Modeling and Visualizing Gathering Structures: An Introduction to VisColl
Alberto Campagnolo and Dot Porter
For our presentation, Dot and Alberto be talking about the VisColl project, which we have been working on for the past five years. VisColl is a system for modeling and visualizing codex manuscripts, and will eventually be expanded to work with early printed books as well.
The principal physical feature of the book in codex format, the gathering structure, is not usually visualized or even mentioned within digitization projects. If this information is recorded at all, it is generally done with the use of collation formulas, a series of letters and numbers that describes how many gatherings there are in a manuscript and how many leaves are in each gathering. There is not a standard schema for manuscript collation formulas and not all practices are able to record accurately the structure of books. VisColl builds on past experiences and through it we strive to describe, visualize, and communicate the gathering structure of books. In our talk we will describe VisColl and provide examples of successful applications of the projects. At the end we’ll describe how future versions will add functionality to link physical details of a manuscript with additional information about the content, which will enable a complete mapping of a physical manuscript.
Alberto Campagnolo, University of Udine, Italy, and Dot Porter, Curator of Digital Research Services at the Schoenberg Institute for Manuscript Studies, have been collaborating on VisColl since 2013.