Fraunhofer analyzes documents faster using artificial intelligence

The Fraunhofer Institute has developed advanced software solutions, DocuLib and NLU.Suite, which leverage artificial intelligence to significantly speed up the digitization, classification, and content analysis of various documents. These solutions utilize deep learning and AI-driven language models, aiming to automate information processing and improve the efficiency of business workflows.

DocuLib is an Optical Character Recognition (OCR) software that employs deep learning models to accurately recognize text, even from documents with poor print quality or aged paper. This enables efficient digitization of physical documents. Subsequently, the NLU.Suite analyzes the content of these digitized documents, or those already in digital format. The system can classify documents, such as invoices or reports, and extract key information like dates and names.

According to the institute, these tools are designed to help companies and public institutions manage escalating document volumes more effectively. They are particularly beneficial for organizations that process large amounts of incoming mail and other business correspondence daily. AI enables automation of tasks like sorting and routing mail to the correct recipients.

Dr. Nicolas Flores-Herr from Fraunhofer IAIS stated the goal is to largely automate information processing through AI, thereby accelerating document-centric workflows. The NLU.Suite utilizes Natural Language Understanding (NLU) aided by language models trained on extensive text data. The combination of statistical methods and rule-based analysis is designed for speed and precision with reduced data requirements.