Analysis of working accident to identify usage drift and contribution of NLP

This communication presents an exploitation of working accidents from the database of occupational accidents EPICEA to identify if hazardous specific situations tied to the use of the working equipment can be highlighted. We present at first the database, its objectives. Then, we clarify the analysis of the narratives of accidents related to particular machines and propose a classification / typology of the conditions of drift use of the equipment which can lead to the accident. Finally, an exploratory application of NLP (Natural Language Processing) tools on 43 accidents is presented.