|本期目录/Table of Contents|

[1]王喆,杨栋梁,况星园,等.考虑提示学习的洪涝灾害应急决策自动问答模型研究*[J].中国安全生产科学技术,2022,18(11):12-18.[doi:10.11731/j.issn.1673-193x.2022.11.002]
 WANG Zhe,YANG Dongliang,KUANG Xingyuan,et al.Research on automatic question answering model of flood disaster emergency decision-making considering Prompt-learning[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2022,18(11):12-18.[doi:10.11731/j.issn.1673-193x.2022.11.002]
点击复制

考虑提示学习的洪涝灾害应急决策自动问答模型研究*
分享到:

《中国安全生产科学技术》[ISSN:1673-193X/CN:11-5335/TB]

卷:
18
期数:
2022年11期
页码:
12-18
栏目:
学术论著
出版日期:
2022-11-30

文章信息/Info

Title:
Research on automatic question answering model of flood disaster emergency decision-making considering Prompt-learning
文章编号:
1673-193X(2022)-11-0012-07
作者:
王喆杨栋梁况星园刘丹马勇
(1.武汉理工大学 安全科学与应急管理学院,湖北 武汉 430070;
2.武汉理工大学 中国应急管理研究中心,湖北 武汉 430070;
3.武汉理工大学 航运学院,湖北 武汉 430063)
Author(s):
WANG Zhe YANG Dongliang KUANG Xingyuan LIU Dan MA Yong
(1.School of Safety Science and Emergency Management,Wuhan University of Technology,Wuhan Hubei 430070,China;
2.China Emergency Management Research Center,Wuhan University of Technology,Wuhan Hubei 430070,China;
3.School of Navigation,Wuhan University of Technology,Wuhan Hubei 430063,China)
关键词:
洪涝灾害应急决策自动问答语言模型提示学习
Keywords:
flood disaster emergency decision-making automatic question answering language model Prompt-learning
分类号:
TP273;X913
DOI:
10.11731/j.issn.1673-193x.2022.11.002
文献标志码:
A
摘要:
为提高洪涝灾害应急处置时效性和科学性,构建洪涝灾害应急决策自动问答系统模型,以提高应急指挥团队的决策效率,在分析洪涝灾害应急决策逻辑基础上,以摘要式问答为任务框架,收集整理包含洪涝灾害应急情景和应急决策的摘要式问答对数据集,建立可用于问答生成的GPT2预训练语言模型,并引入提示学习(Prompt-learning),通过自动创建连续型前缀提示(Prompt),优化少量连续参数,缓解问答对数据较少带来的过拟合风险,利用人工评估和自动评估2种方法验证模型的有效性。研究结果表明:通过GPT2与提示学习相结合建立的自动问答模型,可根据洪涝灾害情景生成语言质量良好及决策信息丰富的答案,有利于提高洪涝灾害应急处置中的科学决策能力。
Abstract:
To improve the timeliness and scientificity of emergency response to the flood disasters,an automatic question answering system model for the emergency decision-making of flood disasters was constructed to improve the decision-making efficiency of emergency command team.Based on the analysis on the emergency decision-making logic of flood disasters,the abstractive question answering was taken as the task framework,and the data sets of abstractive question answering containing the emergency scenarios and emergency decision-making of flood disasters were collected and sorted out.Then a GPT2 pre-training language model that could be used for the question answering generation was established,and the Prompt-learning was introduced.By automatically creating the continuous prefix Prompt,a small number of continuous parameters were optimized to alleviate the risk of over-fitting caused by less data in question answering.Two methods of human evaluation and automatic evaluation were used to verify the effectiveness of the model.The results showed that the automatic question answering model established by the GPT2 and Prompt-learning could generate the answer with good language quality and rich decision-making information according to the flood disaster scenarios.It is beneficial to improve the scientific decision-making ability in the emergency response of flood disasters.

参考文献/References:

[1]中华人民共和国应急管理部.应急管理部发布 2021年全国自然灾害基本情况[EB/OL].(2022-01-23)[2022-05-13].https://www.mem.gov.cn/xw/yjglbgzdt/202201/t20220123_407204.shtml.
[2]郝阳,邓云峰,程五一.基于“情景-应对”数据的高层官员应急决策行为分析框架[J].中国安全生产科学技术,2019,15(3):5-10. HAO Yang,DENG Yunfeng,CHENG Wuyi,et al.Analysis framework for emergency decision-making behavior of senior officials based on scenario-response data[J].Journal of Safety Science and Technology,2019,15(3):5-10.
[3]何帔雨,李鹏,谢汝生,等.基于遗传算法的输电线路覆冰灾害应急响应物资储备决策优化模型[J].中国安全生产科学技术,2019,15(1):51-55. HE Peiyu,LI Peng,XIE Rusheng,et al.Optimization model on decision-making of emergency response material reserve for icing disaster of power transmission line based on genetic algorithm[J].Journal of Safety Science and Technology,2019,15(1):51-55.
[4]王喆,王世昌,涂圣友,等.应急行动方案决策研究综述[J].武汉理工大学学报(信息与管理工程版),2019,41(5):467-472. WANG Zhe,WANG Shichang,TU Shengyou,et al.Review on emergency incident action plans decision-making research[J].Journal of Wuhan University of Technology(Information & Amp Management Engineering),2019,41(5):467-472.
[5]赵芸,刘德喜,万常选,等.检索式自动问答研究综述[J].计算机学报,2021,44(6):1214-1232. ZHAO Yun,LIU Dexi,WAN Changxuan,et al.A review of research on retrieval automatic question answering[J].Chinese Journal of Computers,2021,44(6):1214-1232.
[6]王喆,蒋壮,王世昌,等.应急智能规划中基于约束满足的资源协作方法[J].系统工程学报,2020,35(6):816-823. WANG Zhe,JIANG Zhuang,WANG Shichang,et al.Resource cooperation method based on constrain satisfaction in emergency intelligent planning[J].Journal of Systems Engineering,2020,35(6):816-823.
[7]王兵,郑亚梅,陈茂柯,等.基于Tri-BiLSTM-CNN的钻井安全问答系统[J].西南石油大学学报(自然科学版),2020,42(6):157-164. WANG Bing,ZHENG Yamei,CHEN Maoke,et al.Question answering system for drilling safety based on Tri-BiLSTM-CNN[J].Journal of Southwest Petroleum University(Science & Technology Edition),2020,42(6):157-164.
[8]陈瑛,张晓强,陈昂轩,等.基于信息抽取的食品安全事件自动问答系统方法研究[J].农业机械学报,2020,51(S2):442-448. CHEN Ying,ZHANG Xiaoqiang,CHEN Angxuan,et al.QA system for food safety events based on information extraction[J].Transactions of the Chinese Society for Agricultural Machinery,2020,51(S2):442-448.
[9]CHAN H Y,TSAI M H.Question-answering dialogue system for emergency operations[J].International Journal of Disaster Risk Reduction,2019,41:101313.
[10]KOCISKY T,SCHWARZ J,BLUNSOM P,et al.The narrativeqa reading comprehension challenge[J].Transactions of the Association for Computational Linguistics,2018(6):317-328.
[11]FRERMANN L.Extractive narrative QA with heuristic pre-training[C]// Proceedings of the 2nd Workshop on Machine Reading for Question Answering,2019.
[12]钱锦,黄荣涛,邹博伟,等.基于多任务学习的生成式阅读理解[J].中文信息学报,2021,35(12):103-111. QIAN Jin,HUANG Rongtao,ZOU Bowei,et al.Generative reading comprehension via multi-task learning[J].Journal of Chinese Information Processing,2021,35(12):103-111.
[13]PANDYA H A,BHATT B S.Question answering survey:directions,challenges,datasets,evaluation matrices [J].Journal of Xidian University,2021,15(4):152-168.
[14]LIU P,YUAN W,FU J,et al.Pre-train,prompt,and predict:a systematic survey of prompting methods in natural language processing[EB/OL].(2021-07-28)[2022-05-13].http://pretrain.nlpedia.ai/ ArXiv Preprint arXiv:2107.13586,2021.
[15]DANIEL K,SEWON M,TUSHAR K,et al.UNIFIEDQA:crossing format boundaries with a single QA system[C]// Findings of the Association for Computational Linguistics:EMNLP 2020,2020:1896-1907.
[16]BROWN T,MANN B,RYDER N,et al.Language models are few-shot learners[C]// Advances in Neural Information Processing Systems 33,2020:1877-1901.
[17]LI X L,LIANG P.Prefix-tuning:optimizing continuous prompts for generation[C]// Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing,2021:4582-4597.

相似文献/References:

[1]胡艳菊,李彦海,胡忆沩.危险化学品突发泄漏事故应急决策系统的开发与应用[J].中国安全生产科学技术,2010,6(4):109.
 HU Yan-ju,LI Yan-hai,HU Yi-wei.Development and application of the emergency decision system for dangerous chemicals burst leaking accident[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2010,6(11):109.
[2]徐亚博,汪彤,王培怡,等.基于案例推理的地铁非常规突发事件应急决策方法研究[J].中国安全生产科学技术,2013,9(8):44.[doi:10.11731/j.issn.1673-193x.2013.08.008]
 XU Ya bo,WANG Tong,WANG Pei yi,et al.Research on emergency decision method of unconventional emergency in subway based on casebased reasoning[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2013,9(11):44.[doi:10.11731/j.issn.1673-193x.2013.08.008]
[3]方来华,闫露璐,吴宗之,等.液碳生产企业安全监控与自控系统开发[J].中国安全生产科学技术,2013,9(9):116.[doi:10.11731/j.issn.1673-193x.2013.09.022]
 FANG Lai hua,YAN Lu lu,WU Zong zhi,et al.Development of safety monitoring and automatic control system for liquid carbon production plant[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2013,9(11):116.[doi:10.11731/j.issn.1673-193x.2013.09.022]
[4]谭睿璞,张文德,陈龙龙.基于单值中智集VIKOR的应急群体决策方法[J].中国安全生产科学技术,2017,13(2):79.[doi:10.11731/j.issn.1673-193x.2017.02.014]
 TAN Ruipu,ZHANG Wende,CHEN Longlong.Study on emergency group decision making method based on VIKOR with single valued neutrosophic sets[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2017,13(11):79.[doi:10.11731/j.issn.1673-193x.2017.02.014]
[5]张忠义,庄越.试论我国城市应急软能力提升路径——基于2016年武汉洪涝灾害的理性思考[J].中国安全生产科学技术,2017,13(3):119.[doi:10.11731/j.issn.1673-193x.2017.03.019]
 ZHANG Zhongyi,ZHUANG Yue.Discussion on path of improving urban emergency soft capacity in China—based on rational thinking of Wuhan flood disaster in 2016[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2017,13(11):119.[doi:10.11731/j.issn.1673-193x.2017.03.019]
[6]温念慈,倪少权,陈钉均,等.城市轨道交通突发大客流协同应急决策研究[J].中国安全生产科学技术,2017,13(7):48.[doi:10.11731/j.issn.1673-193x.2017.07.008]
 WEN Nianci,NI Shaoquan,CHEN Dingjun,et al.Study on collaborative emergency decision of outburst mass passenger flow in urban rail transit[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2017,13(11):48.[doi:10.11731/j.issn.1673-193x.2017.07.008]
[7]马文笑,王德鲁.基于案例推理的突发环境事件应急决策模型[J].中国安全生产科学技术,2017,13(12):85.[doi:10.11731/j.issn.1673-193x.2017.12.013]
 MA Wenxiao,WANG Delu.Emergency decision-making model of emergency environmental accidents based on case-based reasoning[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2017,13(11):85.[doi:10.11731/j.issn.1673-193x.2017.12.013]
[8]高平,夏登友,周扬.基于模糊逻辑理论的应急决策冲突程度表示方法研究[J].中国安全生产科学技术,2018,14(12):158.[doi:10.11731/j.issn.1673-193x.2018.12.026]
 GAO Ping,XIA Dengyou,ZHOU Yang.Research on representation method for conflict degree of emergency decisionmaking based on fuzzy logic theory[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2018,14(11):158.[doi:10.11731/j.issn.1673-193x.2018.12.026]
[9]孙开畅,陈璇,时训先,等.基于区间直觉模糊信息TODM法的水利工程应急决策研究[J].中国安全生产科学技术,2019,15(3):19.[doi:10.11731/j.issn.1673-193x.2019.03.003]
 SUN Kaichang,CHEN Xuan,SHI Xunxian,et al.Study on emergency decisionmaking of hydraulic engineering based on interval intuitionistic fuzzy information TODIM method[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2019,15(11):19.[doi:10.11731/j.issn.1673-193x.2019.03.003]
[10]王永明.突发事件应急决策模拟演练设计理论框架与技术要点[J].中国安全生产科学技术,2019,15(5):5.[doi:10.11731/j.issn.1673-193x.2019.05.001]
 WANG Yongming.Theoretical framework and technical essentials of simulation drill design on emergency decisionmaking of emergencies[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2019,15(11):5.[doi:10.11731/j.issn.1673-193x.2019.05.001]

备注/Memo

备注/Memo:
收稿日期: 2022-05-17
* 基金项目: 教育部人文社会科学研究青年基金项目(20YJC630154);国家自然科学基金项目(62073251,52022073,71501151);湖北省自然科学基金项目(2016CFB467)
作者简介: 王喆,博士,副教授,主要研究方向为应急管理、应急指挥决策、人工智能。
通信作者: 杨栋梁,硕士研究生,主要研究方向为应急管理、人工智能。
更新日期/Last Update: 2022-12-11