Research On Event And Evolution Relationship Extraction For The Construction Of Emergency Semantic Knowledge Base

Emergency is unexpected and uncontrollable,often causing serious disaster consequences and posing a great threat to people's lives and property.Although the emergency plan can guide the emergency rescue from the macro level,it cannot assist the experts to make rescue decisions for specific emergencies from a micro perspective,and the existing artificially established emergency knowledge base lacks further information extraction and Expansion.Therefore,combining the knowledge of security science with computer technology,constructing the Semantic Knowledge Base for Emergency,and further enriching the corpus and improving the semantic information,is conducive to improving the efficiency of emergency decision-making.Experts and scholars in the field of emergency decision-making provide support for semantic information data.Firstly building the basic E-SKB based on the semi-structured data according to the associated open data rules.For the insufficient extraction of event elements when identifying existing repetitive events for existing clustering algorithms,the hierarchical blocking method is used to extract the events from different news websites to increase the event instances and identify the repeated bursts to avoid data redundancy.In order to dig out the entity relationships contained in emergencies to enrich the semantic information of the knowledge base,and to solve the problem of missing information in the study of causality extraction in existing emergencies,combined with the modeling of security science and based on the attention based bidirectional LSTM(Long Short-Term Memory)model extracts the evolutionary relationship and introduces the attention mechanism to highlight the importance of the key words to improve the accuracy of the extraction results.The experimental results show that the proposed method can reduce the number of document comparisons and improve the extraction efficiency and avoid event redundancy.The completeness and accuracy of evolution relationship extraction can be further improved.It provides technical support for the information extraction link constructed by E-SKB,and provides more effective information for emergency response decision making.