215 Ontology-based network graph analysis for Japanese consumer product narrative injury data

<h3>Background</h3> In Japan, the National Institute of Technology and Evaluation (NITE) has done a lot of work collecting consumer product injury data to prevent accidents. However, a wealth of narrative data is still available within the collection system, which needs to be interpreted by safety experts. <h3>Objective</h3> This paper describes the initial approach that combined ontology with the network analysis method to quickly and efficiently provide injury prevention experiences for data learning. <h3>Methods</h3> A sample of 1221 carrying device-related injuries in preschool children from the NITE collection system was used in the study. Based on injury process description ontology (IPD-Onto), the narrative injury data were represented as a graph with the safety expert’s interpretation in a machine-readable format. An open graph platform, Gephi, was used to represent and understand injury processes, network graph analysis and visualization. <h3>Results</h3> The results of the analysis show that the ontology-based network graph analysis could give a better understanding of the injury mechanism, which helps quickly identify the essential nodes and bridges in different risk scenarios. <h3>Conclusions</h3> This provides a computer-assisted processing bridge between narrative injury data and graph analysis methods. It also demonstrates a possible approach to injury prevention decision-making from the perspective of safety theory, including the perspective of the Safety-I and Safety-II.