Multimodal LLMs for Historical Dataset Construction from Archival Image Scans: German Patents (1877-1918)

Niclas Griesshaber, Jochen Streb

We leverage multimodal large language models (LLMs) to construct a dataset of 306,070 German patents (1877-1918) from 9,562 archival image scans using our LLM-based pipeline powere…