Creation of a Safety Information Platform Applying Al Analysis Technology: Prototyping Analysis Methods

Industrial plant operators have so far relied mainly on the wisdom and experience of experts to make effective use of huge amounts of text information such as accident cases and near-miss reports. However, the aging and retirement of experienced skilled workers has made continued effective use into a common issue for industrial plant operators. The present study created prototypes of two analysis methods, analysis based on words (text mining + Bayesian network) and analysis based on technical data (ontology) as possible solutions to this problem. In the future, we plan to systematize the created prototype for use in the cloud, and to build a "safety information platform" equipped with a "safety information database" and "analysis software." In addition, we will also con-tinue to improve the operation/management method after the system is launched for connection to a system that allows continuous operation.