Researchers at Teklia are involved in national and international research projects in historical document analysis and understanding.
Census
Census are a major source of information for historical studies. They provide information on the complete population of the country over large periods of time. TEKLIA is a partner of the research project SOCFACE and will provide the complete automatic transcription of the handwritten lists of all the French census between 1836 and 1936.
Norwegian correspondence
The National Library of Norway (NLN) holds a large collection of letters and manuscripts. However, the heterogeneity of handwritten documents is a bigger challenge in handwritten text recognition (HTR) than in OCR. The goal of the Hugin-Munin project is to extend the HTR capabilities to make handwritten documents searchable with high accuracy on the NLN portal. To reach this goal, TEKLIA will develop high performance HTR for Norwegian based on the latest deep learning techniques.
Retro-conversion of Finding Aids
In order to accelerate the online publication of the inventories, the National Archives chose TEKLIA to develop an application to assist in the retro-conversion of handwritten finding aids. This application relied on TEKLIA's expertise in handwriting recognition to automate a large part of the image transcription.
For more information:
- read our blog post : SIMARA: automatic conversion of finding aids with handwriting recognition
- see our research paper : SIMARA: a database for key-value information extraction from full-page handwritten documents
B. Bruyère notebooks
The French Institute for Oriental Archaeology (IFAO) is home to the excavation notebooks of the French Egyptologist Bernard Bruyère (1879-1971). The IFAO commissioned TEKLIA to automatically transcribe these notebooks in order to perform an automatic indexation, making them more readily exploitable by researchers.
Natural history collections
Within the framework of the European project Synthesys+, TEKLIA has developed in collaboration with the Natural History Museum of London a processing chain allowing the automatic recognition of information related to digitised specimens.
Quebec parish records
Québec's parish registers, held in the Bibliothèque et Archives nationales du Québec (BAnQ), have for more than 40 years been manually transcribed and processed by the researchers of the Balsac project. In order to speed up the processing of baptism, marriage and death records for the period 1850-1920, TEKLIA was tasked with developing a complete handwriting recognition and information extraction system.
Books of Hours
Books of hours, with more than 5,000 witnesses preserved throughout the world, form a crucial set for understanding the medieval mental universe. With the HORAE project, TEKLIA has developed, in collaboration with Dominique Stutzmann of the Institut de la recherche et d'histoires des textes (IRHT), a document analysis and text recognition process adapted to books of hours.
Charters and Cartularies
Within the framework of the European HIMANIS project, the whole 80,000 pages of the medieval registers of the French royal chancery were automatically transcribed and indexed for the first time. Following the success of HIMANIS, the European HOME project will extend the automatic transcription to other European medieval manuscripts.