Applying Natural Language Processing in Manufacturing
Marvin Carl May, Jan Neidhöfer, Tom Körner, Louis Schäfer, Gisela Lanza
Abstract
Despite great progress in the digitization of the industrial sector through Industry 4.0 and widely available data, data analysis is typically constrained to numerical data, not the synthesis of knowledge. Although valuable employee knowledge in manufacturing is often described textually, it is rarely formalized and effective application hindered. To close this gap, we introduce methods of Natural Language Processing (NLP) to leverage available text data in manufacturing. For this purpose, we develop a NLP pipeline to handle textual information from machine providers. We extend this with production specific information to reduce failure downtime. The resulting, formalized knowledge can furthermore be used as a basis for optimizing manufacturing processes.