Assurez-vous que votre profil candidat soit correctement renseigné avant de postuler. Les informations de votre profil complètent celles associées à chaque candidature. Afin d’augmenter votre visibilité sur notre Portail Emploi et ainsi permettre aux recruteurs de consulter votre profil candidat, vous avez la possibilité de déposer votre CV dans notre CVThèque en un clic !
General information
Reference : UPR841-DOMSTU-008
Workplace : AUBERVILLIERS
Date of publication : Friday, June 3, 2022
Scientific Responsible name : Dominique STUTZMANN (UPR 841 IRHT) / Mathieu AUBRY (UMR8049 LIGM)
Type of Contract : PhD Student contract / Thesis offer
Contract Period : 36 months
Start date of the thesis : 1 October 2022
Proportion of work : Full time
Remuneration : 2 135,00 € gross monthly
Description of the thesis topic
Features of medieval scripts (CrEMe “Caractérisation des écritures médiévales” research project).
The PhD candidate will be primarily a humanities student and will have to develop a connection to both project teams (IRHT and LIGM), as well as to two scholarly fields, medieval history and the study of scripts, on the one hand, and computer vision and machine learning, on the other.
As their main scholarly target, the PhD candidate will characterise the diverse Latin scripts of the Middle Ages and study their evolution mechanisms based on the morphology of their signs and letter forms.
They will first adapt existing tools to create “prototypes-based descriptions”, i.e. an average shape and a description of its variations, of the letter forms, and validate their output against palaeographical literature. Based on discussions on the relevance of corpora and script types and applicability of techniques with both supervisors, they will be able to choose their corpora according to their previous expertise or fields of interest. Synchronic studies and comparisons may be relevant (e.g. textualis, cursiva, hybrida, semitextualis), as well as diachronic ones (uncial, semi-uncial, Caroline, praegothica). Depending on the results in this first phase, an additional annotation phase may be required to create valid prototypes. The scope of additional annotations will be informed by the palaeographical literature and the analysis of existing hyperdiplomatic editions, which have identified relevant phenomena to analyse scripts.
In a further development, the PhD candidate will propose a strategy to apply and develop the relevant tools and exploit their outputs by creating prototypes and measuring the formal transformations, based on already available corpora. They will aim at proposing weighted metrics for the palaeographical analysis, integrating the metrics generated by artificial vision (distance between prototypes and between prototypes and their instances) and existing criteria of palaeographical analysis.
From then on, the PhD candidate will extend their study to still unannotated corpora in order to be able to characterise and classify the diverse script types at the scale of the medieval millennium.
The PhD candidate will have the opportunity and required scholarly freedom to devise their own experiments and transfer their colleagues’ outputs to their own research.
Additional information: https://oriflamms.hypotheses.org/1885
Work Context
The PhD position is part of the CrEMe Caractérisation des écritures médiévales research project which gathers two CNRS labs in an interdisciplinary research: Institut de Recherche et d’Histoire des Textes (UPR841) and Laboratoire d’informatique Gaspard Monge (UMR8049).
The project is funded by the MITI (Mission for Transverse and Interdisciplinary Initiatives).
CrEMe gathers researchers and teams pertaining to INSHS (Institute for Humanities and Social Sciences) and INS2I (Institute for Information Sciences) for an interdisciplinary study on handwritten scripts of the Middle Ages.
We will study the evolution of Latin scripts and their mechanisms thanks to deep learning and its application to characterize the features of handwritten artifacts, in a multilingual environment, as well at the level of individual scribes as at the collective level and on a long period. We wish to understand how script types emerge, are canonised, and how script families influence each other. With this aim, we will isolate signs and letter forms as if they were iconographic motives.
In the Humanities field, this research will formalize applicable analysis criteria and thus modify our understanding of script history (script classification, dating, evolution process, morphology and formality). For Information Sciences, this research will contribute to enhance the unsupervised learning techniques of neural networks, so as to produce traceable and interpretable decisions – a major challenge in developing deep learning capabilities well beyond this project.
Constraints and risks
Screen work
Additional Information
Knowledge and Skills
– Solid knowledge on the history medieval book scripts, ability to read them (including cursive instances).
– Basic knowledge of Latin and middle French
– Programming languages: Python ; optional : R.
Source : CNRS