Reasearch On Emergency Decision Support Method Of Emergencies Based On Similarity Calculation

Recently, all types of emergencies occur frequently with changes in technological development, climate and so on, it causes serious casualties and bad influence. Therefore, research on emergency management has urgent practical needs. Through analyzing the past emergencies case, experiences and lessons learned can assist decision makers in the face of a new unexpected event to make better decisions. Certainly, the more similar of a case, the greater reference value it has. So, how to find the most similar emergency from the library has become a key issue, it raises the study about similarity of emergencies in this thesis.Multiple incidents are expressed by text, text similarity calculation has had a mature development. Currently, classic algorithms are based on vector space model or longest common subsequence model. However, most of existed research make ordinary text as the target, they use methods based on word sequences, keyword frequency or semantic structure, similarity calculation for emergencies text has not been studied.In assessing the similarity between the ordinary text, main consideration are semantics and syntax. When measuring the similarity of emergencies text, main consideration are state property, cause, decision and other characteristic. Therefore, this thesis proposes a frame based on the characteristics of emergencies, all properties of an emergency will be divided into four kinds:basic properties, disaster bearing carrier, emergency management, secondary derivative disaster. Then this thesis proposes a similarity calculation method based on HowNet and dependency structure analysis, and proposed the core idea that we can measure the similarity of the emergency text by calculating the similarity between frames.Lastly, this thesis proposes a decision support method of emergencies based on similarity calculation. Overall functional processes are:Firstly, extract information of the event using dependency patterns; Then calculate the similarity between the new frame and the frames of previous cases, find out the most similar case from the past. So we can utilize the experience of the past case to assist this new emergency decision.