Construction and Analysis of the Event Evolution Graph for Mass Emergency Based on the Large Language Model
[Research purpose]Based on large language model,an Event Evolution Graph for Mass Emergency is constructed.Through the visual analysis of the time evolution path and the relational evolution path,the development law of mass emergency is revealed.The construction and analysis of event evolution diagrams provide theoretical and practical references for government emergency decision-mak-ing.[Research method]The ontology model of mass emergency is constructed,and the multi-round prompt template based on ChatGPT is designed to realize zero-shot information extraction.The evaluation criteria are designed to evaluate the effect of large model information extraction objectively.The Neo4j database is used to construct and visualize the event evolution graph.And then the evolution analysis of mass emergency is carried out by querying event evolution graph.[Research conclusion]The scheme for information extraction of mass emergency based on large language model not only gives play to the intelligence of contextual language learning of large language model,but also saves time and computational resources.The F1 value of each task is better than that of traditional deep learning model,which verifies its feasibility in the automatic construction of event graph and its effectiveness in the evolution analysis of emergency.It provides a new direction for the application of large language model in emergency decision making.
public emergency eventlarge language modelontology modelinginformation extractionevent evolution graphevolu-tionary analysis
万方数据
维普
