CV
Mathieu Rivière
Second-year MA student in “Digital and Computational Humanities” — École nationale des chartes · Université PSL
Email: mathieu.riviere@chartes.psl.eu
Website: https://icimathieu.github.io/
GitHub: https://github.com/icimathieu
Hugging Face: https://huggingface.co/icimathieu
ORCID: https://orcid.org/0009-0006-5829-7152
HAL: idHAL mathieu-riviere-2003
Profile
My history of science thesis focuses on the “pasteurization” of metropolitan France (1870-1914) and brings together Pasteurians, scientific journals, the press, and parliamentarians under the Third Republic; it relies on scraping, OCR, corpus structuring, annotation, RAG, and classification.
Outlook: the agrégation in history, then a PhD with a strong digital and computational component.
Education
-
École nationale des chartes, École normale supérieure & Université PSL — MA in “Digital and Computational Humanities” (2025 – present) Research seminars in history and philosophy (EHESS, ENS); practical digital humanities classes (ENC); geography lectures (Sorbonne); PSL evening course “A History of AI (1940-2026)” (M. Husson, 2026-2027). Designed the “Alumni” page of the master’s website (hn.chartes.psl.eu). Member of the CartaData junior enterprise, involved in various client projects.
-
Université Paris 1 Panthéon-Sorbonne — BA in History, highest honours (2022 – 2025) Additional courses in history and computer science; history and philosophy seminars; auditing political science courses; evening classes at the Louvre.
-
Lycée Saint-Louis (Paris) — CPGE PCSI, preparatory class in physics, chemistry and engineering sciences (2021 – 2022) — ranked 23rd / 46.
Research experience
Research internship — named-entity recognition (NER)
May – December 2026 — ANR project DECIDON (WP3) — EPITA · ENC · EHESS (LaDéHiS) A computational study of parliamentary life under the Third Republic (1870-1940): how public problems circulate between the press and Parliament, and agenda-setting mechanisms. Annotation guide and schema, coordination of the annotation effort; benchmark of six families of NER approaches (rules, spaCy, BERT, GLiNER, LLMs, ensembles); voting system selected for deployment on the 1881-1940 debates; relation extraction. OCR and structuring of the Journal officiel name indexes (1881-1939). Participation, with the DECIDON team, in the Impresso@BnF datathon (28-30 Oct. 2026). Equivalent to 5 months full-time.
Internship — digital editing and textometry
January – April 2025 — CACTUS group, ENS de Lyon Participation in the digital edition of medieval texts of the Base de Français Médiéval: metadata research and integration; manuscript/text alignment; proofreading of editions; zone annotation (bounding boxes) on illuminations.
Digital humanities projects (selection)
Thesis — building a corpus on “pasteurization” (1870-1914) (2025 – present)
History of science thesis, supervised by Anne Rasmussen (EHESS) and Marie Puren (EPITA, ENC). Gallica scraping and metadata extraction; OCR of scientific journals (late 19th – early 20th century) with a benchmark of several models; corpus and database structuring.
- Repositories: scraping_pdf · transcription · Mini-thesis (PDF, in French): latex_memoire
“Structured Data” and “Spatialization” workshops — co-facilitation (June 2026)
From Closure to Sharing summer school (heritage of Afghanistan and neighbouring territories), INHA & Campus Condorcet, Huma-Num Distam consortium. With Maxime Létoffé (ENC, EPHE), in English: spatialization of the IFRI photographic collection (cards in Persian/Latin) — OCR, constrained LLM structuring, Wikidata linking, GeoNames geocoding. CartaData assignment.
- Repository: atelier_spatialisation_INHA
Vaucluse postcards — VLM pipeline & mapping (January – April 2026)
ENC hackathon and CartaData junior enterprise. Postcard OCR with VLMs, metadata extraction and cleaning, database querying via API as a safeguard, georeferencing, JSON/CSV structuring and interactive map (HTML). Pipeline being generalized to other archival collections. Round table with the archivists and IT department of the Archives de Vaucluse (Avignon, 16 Feb. 2026); lightning talk and demo at the ADEMEC day (2 Dec. 2026).
- Map published by the Département: maps.vaucluse.fr · Repository · Demo
Technical skills
- Programming & data: Python (numpy, pandas, networkx, scikit-learn, PyTorch); HTML, JavaScript, RSS.
- AI: LLMs and VLMs (Gemma, Qwen, Mistral, Gemini; via API or locally with vLLM, llama.cpp); prompting, constrained decoding, agents; fine-tuning (QLoRA), distillation; synthetic data generation (augmentation, degradation); knowledge graphs and GNNs.
- NLP & textometry: TXM, Iramuteq; spaCy, BERT models (CamemBERT, hmBERT), GLiNER; topic modeling; classification, NER, relation extraction, RAG, stylometry, emotion detection.
- OCR, computer vision & annotation: pero-ocr, YOLO (local); Qwen-VL, Gemini (API); Label Studio; design of annotation schemas and guides.
- Spatialization: geocoding (Nominatim, GeoNames, Wikidata), georeferencing, structuring of geographic metadata, geopandas, interactive mapping (folium); LLM agents coupled with a search engine; street-level photo geolocation through image matching (MegaLoc, LightGlue); 3D reconstruction (LiDAR HD).
- Scraping: BeautifulSoup (HTML parsing), Selenium (anti-bot bypass), Scrapy (crawling) — press and archives (L’Histoire, Le Monde diplomatique, Archives de Vaucluse); IIIF and heritage APIs (Gallica, BnF).
- Structuring & databases: XML-TEI, XPath, SQL, LaTeX; digital editing workflows.
- Tools: Git/GitHub, Hugging Face; operating remote machines and servers (SSH).
Languages
French native · English C1 (TOEIC 860, 2020) · German B1–B2.
Outreach & science communication
Video creator — Histosef
2023 – 2026 Popular history videos on YouTube (in French), based on scholarly readings and essays.
Writer — L’Ouvreuse (Sorbonne webzine)
2024 – 2025