Personalized Safety Instruction Recommendation for construction workers Based on Apriori Algorithm

The safety management is a focus in the construction industry, and accident causal factors are various. Many management methods are applied to improve the safety state and safety instruction is an essential part of this. However, traditional safety instruction method is always not personalized and wastes the workers more time. The authors improved the safety instruction by analysing real accident cases and the rule-based recommendation method. First, the authors analyzed 157 accident cases by Systematic Knowledge Engineering method, and and found safety accident cases mainly contain six kinds of information: accident expense, accident types, accident time, causal factors, accident location, and worker types. The causal factors are the most important part of case analysis, and they can be divided into six sub-causes according to literature review: management level, input resource to safety, construction hazards, equipment and materials, weather, qualification, physical condition, and mental state. Then Apriori algorithm is used to explore the association among these accident factors to recommend more precise safety instruction for workers. Finally, expert workshops are held to evaluate the analysis results of Apriori algorithm. Preliminary results show that the model can improve the traditional safety instruction by providing more precise relationship among the worker characterises and safety accidents.