Data-knowledge Driven Method For Mining Emergency Response Rules

When an emergency occurs,it is very important to provide an effective response to the situation in time to reduce the harm of the accident.Historical emergency cases contain a wealth of effective response experience and knowledge.Based on the historical emergency case set,through the use of association rules mining method,we can find the fine-grained association rules reflected in the case,considering the relationship between some scenarios and response activities.This kind of association rules,such as "if scenarios then response ",as an empirical knowledge,can not only provide reference for the refinement and improvement of emergency plans,but also assist decision-makers to make emergency decisions in some of the same or similar situations.However,the existing research has the following limitations,On the one hand,in the process of pattern mining,the existing research only considers the support index and ignores the correlation degree between the items in the pattern,resulting in a large number of uncorrelated frequent or rare patterns.On the other hand,most of the existing researches focus on frequent association rules mining,ignoring the importance of rare association rules mining,and lack of ideas and frameworks covering frequent and rare association rules mining;In addition,the existing research lacks the guidance of domain knowledge,resulting in a large number of frequent and rare patterns that have nothing to do with coping,which makes it difficult for decision makers to quickly obtain valuable information they are concerned about.To overcome the above problems,this paper proposes the corresponding solutions.(1)Aiming at the problem that the existing frequent and rare pattern mining methods only consider the support index and ignore the correlation between the items in the pattern,resulting in a large number of uncorrelated patterns,this paper constructs an improved bond index,which is suitable for the "scenario response" mode of emergency.Furthermore,a method of mining frequent and rare coping patterns based on improved bond index is proposed to reduce the occurrence of uncorrelated frequent and rare coping patterns.(2)Aiming at the lack of domain knowledge guidance in existing methods and ignoring the problem of association rule mining in rare cases,it constructs a research framework for the mining of rare and frequent association rules driven by emergency case data and emergency domain knowledge.Under the guidance of the knowledge element network,the method of mining the association rules of "scenario-response activity" contained in historical emergency cases,including Rare Emergency Response Association Rules mining method based on Improved Bond Index(IBI-RERAR)and Frequent Emergency Response Association Rules mining method based on Knowledge Direction and Improved Bond Index(KDIBI-FERAR),to improve the efficiency and quality of excavation.In this paper,a real case of hazardous chemicals leakage is taken as an example to verify the proposed method.The experimental results show that,on the one hand,the proposed method can reduce the occurrence of uncorrelated frequent and rare emergency response patterns,on the other hand,it can help decision-makers quickly find frequent and rare association rules related to emergency response from emergency cases,and provide support for "scenario response" decision-making of emergencies,which has a good application prospect.