Article(id=1148106728694084402, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1148106708670477182, articleNumber=1003-3033(2025)03-0204-08, orderNo=null, doi=10.16265/j.cnki.issn1003-3033.2025.03.0223, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1728835200000, receivedDateStr=2024-10-14, revisedDate=1734451200000, revisedDateStr=2024-12-18, acceptedDate=null, acceptedDateStr=null, onlineDate=1751659574911, onlineDateStr=2025-07-05, pubDate=1743091200000, pubDateStr=2025-03-28, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1751659574911, onlineIssueDateStr=2025-07-05, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1751659574911, creator=13701087609, updateTime=1751659574911, updator=13701087609, issue=Issue{id=1148106708670477182, tenantId=1146029695717560320, journalId=1146031787341344770, year='2025', volume='35', issue='3', pageStart='1', pageEnd='268', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=0, createTime=1751659570138, creator=13701087609, updateTime=1757401518130, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1172190184155238915, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1148106708670477182, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1172190184155238916, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1148106708670477182, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=204, endPage=211, ext={EN=ArticleExt(id=1149767356206985749, articleId=1148106728694084402, tenantId=1146029695717560320, journalId=1146031787341344770, language=EN, title=Gas knowledge bidirectional encoder representations from transformers model based on knowledge injection, columnId=1149733270084042840, journalTitle=China Safety Science Journal, columnName=Public safety, runingTitle=null, highlight=null, articleAbstract=

In order to enhance emergency management in the field of gas pipeline networks,Gas-kBERT model was proposed. The model incorporated data from the gas pipeline network field expanded by Chat Generative Pre-Trained Transformer,(ChatGPT)and Chinese Gas Language Understanding Subject-Predicate-Object(CGLU-Spo) and related corpora were constructed in this field. By altering the model's masking (MASK) mechanism,domain knowledge was successfully injected into the model. Considering the professionalism and specificity of the gas pipeline network field,Gas-kBERT was pre-trained on various scales and contents of corpora and fine-tuned on named entity recognition and classification tasks within this field. Experimental results demonstrated that,compared to the general BERT model,Gas-kBERT exhibited significant performance improvements in F1-score in text mining tasks in the gas pipeline network field. Specifically,in the named entity recognition task,the F1-score was increased by 29.55%,and in the text classification task,the F1-score improvement reached up to 83.33%. This study proves that the Gas-kBERT model performs exceptionally well in text mining tasks in the gas pipeline network field.

, correspAuthors=Yufeng ZHUANG, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Xiaoyu LIU, Yufeng ZHUANG, Xinghao ZHAO, Kefan WANG, Guokai ZHANG), CN=ArticleExt(id=1148106732204716147, articleId=1148106728694084402, tenantId=1146029695717560320, journalId=1146031787341344770, language=CN, title=基于知识注入的燃气知识双向变换器模型, columnId=1149733271510106222, journalTitle=中国安全科学学报, columnName=公共安全, runingTitle=null, highlight=null, articleAbstract=

为提高燃气管网领域的应急管理水平,提出燃气知识双向变换器(Gas-kBERT)模型。该模型结合聊天生成预训练转换器(ChatGPT)扩充的燃气管网领域数据,以及构建的中文燃气语言理解-三元组(CGLU-Spo)和相关语料库,通过改变模型的掩码(MASK)机制,成功将领域知识注入模型中。考虑到燃气管网领域的专业性和特殊性,Gas-kBERT在不同规模和内容的语料库上进行预训练,并在燃气管网领域的命名实体识别和分类任务上进行微调。结果表明:与通用的双向变换器(BERT)模型相比,Gas-kBERT在燃气管网领域的文本挖掘任务中F1值表现出显著的提升。在命名实体识别任务中,F1值提高29.55%;在文本分类任务中,F1值提升高达83.33%。由此证明Gas-kBERT模型在燃气管网领域的文本挖掘任务中具有出色的表现。

, correspAuthors=庄育锋, authorNote=null, correspAuthorsNote=
** 庄育锋(1972—),男,上海人,博士,教授,主要从事智能控制与装备安全等方面的研究。E-mail:
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=UL8axGWfLZIhpkAsc7GAwQ==, magXml=celGXeTK2FUOi+Y/XmO44w==, pdfUrl=null, pdf=JgO1uGXUk6wIAewOiNll6A==, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=ImfvZIPjJAqIRwxdq6xq0g==, mapNumber=null, authorCompany=null, fund=null, authors=

柳晓昱 (2000—),女,天津人,硕士研究生,主要研究方向为文本挖掘与物流信息技术及工程应用。E-mail:

, authorsList=柳晓昱, 庄育锋, 赵兴昊, 王珂璠, 张国开)}, authors=[Author(id=1165678285049574237, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=ivy@bupt.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1165678285104100191, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285049574237, language=EN, stringName=Xiaoyu LIU, firstName=Xiaoyu, middleName=null, lastName=LIU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1165678285150237536, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285049574237, language=CN, stringName=柳晓昱, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 智能工程与自动化学院,北京 100876, bio={"img":"ye3rgCxZBdwBwwSHGbXQ7Q==","content":"

柳晓昱 (2000—),女,天津人,硕士研究生,主要研究方向为文本挖掘与物流信息技术及工程应用。E-mail:

"}, bioImg=ye3rgCxZBdwBwwSHGbXQ7Q==, bioContent=

柳晓昱 (2000—),女,天津人,硕士研究生,主要研究方向为文本挖掘与物流信息技术及工程应用。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1165678284982465369, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, xref=null, ext=[AuthorCompanyExt(id=1165678284990853978, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China), AuthorCompanyExt(id=1165678284995048283, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 智能工程与自动化学院,北京 100876)])]), Author(id=1165678285200569186, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=zhuangyf@bupt.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1165678285255095140, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285200569186, language=EN, stringName=Yufeng ZHUANG, firstName=Yufeng, middleName=null, lastName=ZHUANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=**, address=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1165678285309621093, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285200569186, language=CN, stringName=庄育锋, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=**, address=北京邮电大学 智能工程与自动化学院,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1165678284982465369, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, xref=null, ext=[AuthorCompanyExt(id=1165678284990853978, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China), AuthorCompanyExt(id=1165678284995048283, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 智能工程与自动化学院,北京 100876)])]), Author(id=1165678285359952743, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1165678285414478697, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285359952743, language=EN, stringName=Xinghao ZHAO, firstName=Xinghao, middleName=null, lastName=ZHAO, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1165678285502559082, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285359952743, language=CN, stringName=赵兴昊, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 智能工程与自动化学院,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1165678284982465369, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, xref=null, ext=[AuthorCompanyExt(id=1165678284990853978, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China), AuthorCompanyExt(id=1165678284995048283, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 智能工程与自动化学院,北京 100876)])]), Author(id=1165678285565473644, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1165678285649359726, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285565473644, language=EN, stringName=Kefan WANG, firstName=Kefan, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1165678285783577455, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285565473644, language=CN, stringName=王珂璠, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 智能工程与自动化学院,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1165678284982465369, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, xref=null, ext=[AuthorCompanyExt(id=1165678284990853978, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China), AuthorCompanyExt(id=1165678284995048283, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 智能工程与自动化学院,北京 100876)])]), Author(id=1165678285884240753, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1165678285951349619, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285884240753, language=EN, stringName=Guokai ZHANG, firstName=Guokai, middleName=null, lastName=ZHANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1165678286010069876, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, authorId=1165678285884240753, language=CN, stringName=张国开, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 智能工程与自动化学院,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1165678284982465369, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, xref=null, ext=[AuthorCompanyExt(id=1165678284990853978, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China), AuthorCompanyExt(id=1165678284995048283, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 智能工程与自动化学院,北京 100876)])])], keywords=[Keyword(id=1165678286119121781, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, orderNo=1, keyword=gas pipeline networks), Keyword(id=1165678286207202166, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, orderNo=2, keyword=gas knowledge bidirectional encoder representations from transformers(Gas-kBERT)model), Keyword(id=1165678286274311031, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, orderNo=3, keyword=natural language processing(NLP)), Keyword(id=1165678286320448376, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, orderNo=4, keyword=knowledge injection), Keyword(id=1165678286366585721, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, orderNo=5, keyword=bidirectional encoder representations from transformers (BERT)), Keyword(id=1165678286421111674, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, orderNo=1, keyword=燃气管网), Keyword(id=1165678286488220541, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, orderNo=2, keyword=燃气知识双向变换器(Gas-kBERT)模型), Keyword(id=1165678286572106625, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, orderNo=3, keyword=自然语言处理(NLP)), Keyword(id=1165678286622438276, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, orderNo=4, keyword=知识注入), Keyword(id=1165678286676964231, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, orderNo=5, keyword=双向变换器(BERT)模型)], refs=[Reference(id=1165678288123999138, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=1996, volume=51, issue=1/3, pageStart=77, pageEnd=92, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=MONTIEL H, VÍLCHEZ JA, ARNALDOS J, journalName=Journal of Hazardous Materials, refType=null, unstructuredReference=MONTIEL H, VÍLCHEZ JA, ARNALDOS J, et al. Historical analysis of accidents in the transportation of natural gas[J]. Journal of Hazardous Materials, 1996, 51 (1/3):77-92., articleTitle=Historical analysis of accidents in the transportation of natural gas, refAbstract=null), Reference(id=1165678288178525091, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=1, pageStart=70, pageEnd=72, url=null, language=null, rfNumber=[2], rfOrder=1, authorNames=郭秀军, 刘嘉柠, journalName=网络安全和信息化, refType=null, unstructuredReference=郭秀军, 刘嘉柠. 智能化燃气管网监测与管理[J]. 网络安全和信息化, 2024(1):70-72., articleTitle=智能化燃气管网监测与管理, refAbstract=null), Reference(id=1165678288233051044, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[3], rfOrder=2, authorNames=李兰, journalName=基于视觉检测与图像处理分析技术的燃气管道检测方法研究, refType=null, unstructuredReference=李兰. 基于视觉检测与图像处理分析技术的燃气管道检测方法研究[D]. 南京: 东南大学, 2019., articleTitle=null, refAbstract=null), Reference(id=1165678288291771301, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[3], rfOrder=3, authorNames=LI Lan, journalName=Study on the gas pipeline detection method based on visual inspection and image processing analysis technology, refType=null, unstructuredReference=LI Lan. Study on the gas pipeline detection method based on visual inspection and image processing analysis technology[D]. Nanjing: Southeast University, 2019., articleTitle=null, refAbstract=null), Reference(id=1165678288392434598, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=4, authorNames=苟爽, journalName=基于机器视觉的燃气管道内环境感知技术研究, refType=null, unstructuredReference=苟爽. 基于机器视觉的燃气管道内环境感知技术研究[D]. 昆明: 昆明理工大学, 2018., articleTitle=null, refAbstract=null), Reference(id=1165678288442766247, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=5, authorNames=GOU Shuang, journalName=Research on internal environmental perception technology of gas pipeline based on machine vision, refType=null, unstructuredReference=GOU Shuang. Research on internal environmental perception technology of gas pipeline based on machine vision[D]. Kunming: Kunming University of Science and Technology, 2018., articleTitle=null, refAbstract=null), Reference(id=1165678288493097896, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2021, volume=41, issue=2, pageStart=32, pageEnd=36, url=null, language=null, rfNumber=[5], rfOrder=6, authorNames=周吉祥, 张浩, journalName=煤气与热力, refType=null, unstructuredReference=周吉祥, 张浩. 智能管道建设下的城镇燃气管道完整性管理[J]. 煤气与热力, 2021, 41(2):32-36,46., articleTitle=智能管道建设下的城镇燃气管道完整性管理, refAbstract=null), Reference(id=1165678288543429545, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2021, volume=41, issue=2, pageStart=32, pageEnd=36, url=null, language=null, rfNumber=[5], rfOrder=7, authorNames=ZHOU Jixiang, ZHANG Hao, journalName=Gas & Heat, refType=null, unstructuredReference=ZHOU Jixiang, ZHANG Hao. Integrity management of urban gas pipeline under construction of intelligent pipeline[J]. Gas & Heat, 2021, 41(2):32-36,46., articleTitle=Integrity management of urban gas pipeline under construction of intelligent pipeline, refAbstract=null), Reference(id=1165678288593761194, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2022, volume=41, issue=6, pageStart=812, pageEnd=817, url=null, language=null, rfNumber=[6], rfOrder=8, authorNames=李思洁, 王亚慧, 张子豪, journalName=消防科学与技术, refType=null, unstructuredReference=李思洁, 王亚慧, 张子豪. 燃气输配突发事件应急处置的知识图谱构建[J]. 消防科学与技术, 2022, 41(6):812-817., articleTitle=燃气输配突发事件应急处置的知识图谱构建, refAbstract=null), Reference(id=1165678288644092843, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2022, volume=41, issue=6, pageStart=812, pageEnd=817, url=null, language=null, rfNumber=[6], rfOrder=9, authorNames=LI Sijie, WANG Yahui, ZHANG Zihao, journalName=Fire Science and Technology, refType=null, unstructuredReference=LI Sijie, WANG Yahui, ZHANG Zihao. Construction of knowledge graph for emergency disposal of gas transmission and distribution emergencies[J]. Fire Science and Technology, 2022, 41(6):812-817., articleTitle=Construction of knowledge graph for emergency disposal of gas transmission and distribution emergencies, refAbstract=null), Reference(id=1165678288694424492, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2008, volume=null, issue=null, pageStart=656, pageEnd=null, url=null, language=null, rfNumber=[7], rfOrder=10, authorNames=TIWARY U S, SIDDIQUI T, journalName=Natural language processing and information retrieval, refType=null, unstructuredReference=TIWARY U S, SIDDIQUI T. Natural language processing and information retrieval[M]. New York: Oxford University Press, 2008:656., articleTitle=null, refAbstract=null), Reference(id=1165678288740561837, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[8], rfOrder=11, authorNames=WANG Chenglong, JIANG Feijun, YANG Hongxia, journalName=The 23rd ACM SIGKDD International Conference, refType=null, unstructuredReference=WANG Chenglong, JIANG Feijun, YANG Hongxia. A hybrid framework for text modeling with convolutional RNN[C]. The 23rd ACM SIGKDD International Conference, 2017: DOI:10.1145/3097983.3098140., articleTitle=A hybrid framework for text modeling with convolutional RNN, refAbstract=null), Reference(id=1165678288790893486, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=20, issue=6, pageStart=13, pageEnd=19, url=null, language=null, rfNumber=[9], rfOrder=12, authorNames=QIN Ya, SHEN Guowei, ZHAO Wenbo, journalName=Frontiers of Information Technology and Electronic Engi-neering, refType=null, unstructuredReference=QIN Ya, SHEN Guowei, ZHAO Wenbo, et al. A network security entity recognition method based on feature template and CNN-BiLSTM-recognition method based on feature template and CNN-BiLSTM-CRF[ J]. Frontiers of Information Technology and Electronic Engi-neering, 2019, 20(6):13-19., articleTitle=A network security entity recognition method based on feature template and CNN-BiLSTM-recognition method based on feature template and CNN-BiLSTM-CRF[ J], refAbstract=null), Reference(id=1165678288837030831, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=145, pageEnd=154, url=null, language=null, rfNumber=[10], rfOrder=13, authorNames=RAO N V, KRISHNA B T, journalName=KALYA S, KULKARNI M, SHIVAPRAKASHA K S. Advances in VLSI, Signal Processing, Power Electronics, IoT, Communication and Embedded Systems, refType=null, unstructuredReference=RAO N V, KRISHNA B T. Automatic modulation recognition using machine learning techniques: a review[C]// KALYA S, KULKARNI M, SHIVAPRAKASHA K S. Advances in VLSI, Signal Processing, Power Electronics, IoT, Communication and Embedded Systems. Singapore: Springer Singapore, 2021: 145-154., articleTitle=Automatic modulation recognition using machine learning techniques: a review, refAbstract=null), Reference(id=1165678288883168176, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2018, volume=92, issue=null, pageStart=35, pageEnd=45, url=null, language=null, rfNumber=[11], rfOrder=14, authorNames=MA Zhiliang, CAI Shiyao, MAO Na, journalName=Automation in Construction, refType=null, unstructuredReference=MA Zhiliang, CAI Shiyao, MAO Na, et al. Construction quality management based on a collaborative system using BIM and indoor positioning[J]. Automation in Construction, 2018, 92:35-45., articleTitle=Construction quality management based on a collaborative system using BIM and indoor positioning, refAbstract=null), Reference(id=1165678288946082737, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=2227, pageEnd=2237, url=null, language=null, rfNumber=[12], rfOrder=15, authorNames=PETERS M, NEUMANN M, IYYER M, journalName=The, refType=null, unstructuredReference=PETERS M, NEUMANN M, IYYER M, et al. Deep contextualized word representations[C]. The 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,2018:2227-2237., articleTitle=Deep contextualized word representations, refAbstract=null), Reference(id=1165678289004802994, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[13], rfOrder=16, authorNames=DEVLIN J, CHANG Mingwei, LEE Kenton, journalName=null, refType=null, unstructuredReference=DEVLIN J, CHANG Mingwei, LEE Kenton, et al. BERT: pre-training of deep bidirectional transformers for language understanding[C]. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019.DOI:10.48550/arXiv.1810.04805., articleTitle=BERT: pre-training of deep bidirectional transformers for language understanding, refAbstract=null), Reference(id=1165678289050940339, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=45, issue=1, pageStart=140, pageEnd=147, url=null, language=null, rfNumber=[14], rfOrder=17, authorNames=邓存彬, 虞慧群, 范贵生, journalName=华东理工大学学报:自然科学版, refType=null, unstructuredReference=邓存彬, 虞慧群, 范贵生, 等. 燃气客服热线的中文文本情感分析[J]. 华东理工大学学报:自然科学版, 2019, 45(1):140-147., articleTitle=燃气客服热线的中文文本情感分析, refAbstract=null), Reference(id=1165678289101271988, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=45, issue=1, pageStart=140, pageEnd=147, url=null, language=null, rfNumber=[14], rfOrder=18, authorNames=DENG Cunbin, YU Huiqun, FAN Guisheng, journalName=Journal of East China University of Science and Technology:Natural Science Edition, refType=null, unstructuredReference=DENG Cunbin, YU Huiqun, FAN Guisheng, et al. Sentiment analysis of Chinese texts for gas customer service hotline[J]. Journal of East China University of Science and Technology:Natural Science Edition, 2019, 45(1):140-147., articleTitle=Sentiment analysis of Chinese texts for gas customer service hotline, refAbstract=null), Reference(id=1165678289151603637, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2023, volume=19, issue=3, pageStart=39, pageEnd=45, url=null, language=null, rfNumber=[15], rfOrder=19, authorNames=王明达, 张榜, 吴志生, journalName=中国安全生产科学技术, refType=null, unstructuredReference=王明达, 张榜, 吴志生, 等. 基于强化学习的城镇燃气事故信息抽取方法[J]. 中国安全生产科学技术, 2023, 19(3):39-45., articleTitle=基于强化学习的城镇燃气事故信息抽取方法, refAbstract=null), Reference(id=1165678289197740982, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2023, volume=19, issue=3, pageStart=39, pageEnd=45, url=null, language=null, rfNumber=[15], rfOrder=20, authorNames=WANG Mingda, ZHANG Bang, WU Zhisheng, journalName=Journal of Safety Science and Technology, refType=null, unstructuredReference=WANG Mingda, ZHANG Bang, WU Zhisheng, et al. Information extraction method of urban gas accidents based on reinforcement learning[J]. Journal of Safety Science and Technology, 2023, 19(3):39-45., articleTitle=Information extraction method of urban gas accidents based on reinforcement learning, refAbstract=null), Reference(id=1165678289243878327, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2020, volume=36, issue=4, pageStart=1234, pageEnd=1240, url=null, language=null, rfNumber=[16], rfOrder=21, authorNames=LEE J H, YOON W J, KIM S D, journalName=Bioinformatics, refType=null, unstructuredReference=LEE J H, YOON W J, KIM S D, et al. BioBERT:a pre-trained biomedical language representation model for biomedical text mining[J]. Bioinformatics, 2020, 36(4):1234-1240., articleTitle=BioBERT:a pre-trained biomedical language representation model for biomedical text mining, refAbstract=null), Reference(id=1165678289294209976, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2019, volume=35, issue=10, pageStart=1745, pageEnd=1752, url=null, language=null, rfNumber=[17], rfOrder=22, authorNames=WANG Xuan, ZHANG Xiang, REN Yuhao, journalName=Bioinformatics, refType=null, unstructuredReference=WANG Xuan, ZHANG Xiang, REN Yuhao, et al. Cross-type biomedical named entity recognition with deep multi-task learning[J]. Bioinformatics, 2019, 35(10):1745-1752., articleTitle=Cross-type biomedical named entity recognition with deep multi-task learning, refAbstract=null), Reference(id=1165678289340347321, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2022, volume=32, issue=2, pageStart=115, pageEnd=120, url=null, language=null, rfNumber=[18], rfOrder=23, authorNames=宋敦江, 杨霖, 钟少波, journalName=中国安全科学学报, refType=null, unstructuredReference=宋敦江, 杨霖, 钟少波. 基于 BERT 的灾害三元组信息抽取优化研究[J]. 中国安全科学学报, 2022, 32(2) : 115-120., articleTitle=基于 BERT 的灾害三元组信息抽取优化研究, refAbstract=null), Reference(id=1165678289386484666, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2022, volume=32, issue=2, pageStart=115, pageEnd=120, url=null, language=null, rfNumber=[18], rfOrder=24, authorNames=SONG Dunjiang, YANG Lin, ZHONG Shaobo, journalName=China Safety Science Journal, refType=null, unstructuredReference=SONG Dunjiang, YANG Lin, ZHONG Shaobo. Research on optimization of disaster triplet information extraction based on BERT[J]. China Safety Science Journal, 2022, 32(2): 115-120., articleTitle=Research on optimization of disaster triplet information extraction based on BERT, refAbstract=null), Reference(id=1165678289428427707, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2023, volume=33, issue=7, pageStart=190, pageEnd=195, url=null, language=null, rfNumber=[19], rfOrder=25, authorNames=郑彬彬, 冯婷婷, 王佳贺, journalName=中国安全科学学报, refType=null, unstructuredReference=郑彬彬, 冯婷婷, 王佳贺, 等. 基于文本挖掘的城镇燃气事故致因及关联分析[J]. 中国安全科学学报, 2023, 33(7):190-195., articleTitle=基于文本挖掘的城镇燃气事故致因及关联分析, refAbstract=null), Reference(id=1165678291424916413, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, doi=null, pmid=null, pmcid=null, year=2023, volume=33, issue=7, pageStart=190, pageEnd=195, url=null, language=null, rfNumber=[19], rfOrder=26, authorNames=ZHENG Binbin, FENG Tingting, WANG Jiahe, journalName=China Safety Science Journal, refType=null, unstructuredReference=ZHENG Binbin, FENG Tingting, WANG Jiahe, et al. Causes and correlation analysis of urban gas accidents based on textmining[J]. China Safety Science Journal, 2023, 33(7): 190-195., articleTitle=Causes and correlation analysis of urban gas accidents based on textmining, refAbstract=null)], funds=[Fund(id=1165678288019141537, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, awardId=52478123, language=CN, fundingSource=国家自然科学基金资助(52478123), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1165678284982465369, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, xref=null, ext=[AuthorCompanyExt(id=1165678284990853978, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China), AuthorCompanyExt(id=1165678284995048283, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, companyId=1165678284982465369, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 智能工程与自动化学院,北京 100876)])], figs=[ArticleFig(id=1165678286840542092, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Fig.1, caption=Gas-kBERT model structure diagram, figureFileSmall=s/b0KnruwTTzWTPTZWP3Dw==, figureFileBig=VwLh3S+yBu+8aZSRch4C5w==, tableContent=null), ArticleFig(id=1165678286907650958, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=图1, caption=Gas-kBERT模型结构, figureFileSmall=s/b0KnruwTTzWTPTZWP3Dw==, figureFileBig=VwLh3S+yBu+8aZSRch4C5w==, tableContent=null), ArticleFig(id=1165678286962176913, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Fig.2, caption=Expanded Text from CGLU-spo, figureFileSmall=CR+fWp+2kH0QXFJXvS4EuQ==, figureFileBig=tRdoR38dxKIBCOGS0ck33w==, tableContent=null), ArticleFig(id=1165678287033480082, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=图2, caption=CGLU-spo扩充后的文本, figureFileSmall=CR+fWp+2kH0QXFJXvS4EuQ==, figureFileBig=tRdoR38dxKIBCOGS0ck33w==, tableContent=null), ArticleFig(id=1165678287079617427, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Fig.3, caption=The masking mechanism of the model, figureFileSmall=tiRJCT3dePvzm0zcGHRFOA==, figureFileBig=TsBwmg0wr5GPoNq+7VtEEw==, tableContent=null), ArticleFig(id=1165678287138337684, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=图3, caption=模型的掩码机制, figureFileSmall=tiRJCT3dePvzm0zcGHRFOA==, figureFileBig=TsBwmg0wr5GPoNq+7VtEEw==, tableContent=null), ArticleFig(id=1165678287192863637, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Fig.4, caption=Gas-kBERT model overall training process, figureFileSmall=PwH6Dzuj7wQ3maC/RYXh2A==, figureFileBig=shKx5XM3x9W8X3yPnoQF0g==, tableContent=null), ArticleFig(id=1165678287243195286, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=图4, caption=Gas-kBERT模型整体训练流程, figureFileSmall=PwH6Dzuj7wQ3maC/RYXh2A==, figureFileBig=shKx5XM3x9W8X3yPnoQF0g==, tableContent=null), ArticleFig(id=1165678287293526935, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Table 1, caption=

Pre-training corpus

, figureFileSmall=null, figureFileBig=null, tableContent=
数据 大小
(原始
文本)
大小
(ChatGPT
增强后)
领域
新闻语料(News_zh_2016) 1.6 G 通用
领域
维基百科(Wiki_zh_2019) 1.2 G 通用
领域
燃气事故报告(Gas Accident Report,GAR) 5.1M 20.3 M 燃气
管网
燃气标准(Gas standards,GS) 3.4M 16.9 M 燃气
管网
燃气新闻(Gas News,GN)、燃气应急预案(Gas Emergency Plan,GEP)等 2.7 M 13.7M 燃气
管网
), ArticleFig(id=1165678287348052888, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=表1, caption=

预训练语料

, figureFileSmall=null, figureFileBig=null, tableContent=
数据 大小
(原始
文本)
大小
(ChatGPT
增强后)
领域
新闻语料(News_zh_2016) 1.6 G 通用
领域
维基百科(Wiki_zh_2019) 1.2 G 通用
领域
燃气事故报告(Gas Accident Report,GAR) 5.1M 20.3 M 燃气
管网
燃气标准(Gas standards,GS) 3.4M 16.9 M 燃气
管网
燃气新闻(Gas News,GN)、燃气应急预案(Gas Emergency Plan,GEP)等 2.7 M 13.7M 燃气
管网
), ArticleFig(id=1165678287410967449, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Table 2, caption=

Pre-training task corpus combination

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 语料组合
BERT News+Wiki
Gas-kBERT(raw) GAR+GS+GN+
GEP(原始文本)
Gas-kBERTv1.0(GAR+GS+
GN+GEP)
GAR+GS+GN+GEP
(ChatGPT增强后文本)
Gas-kBERTv1.1(+GAR+
GS+GN+GEP)
News+Wiki+GAR+GS+GN+
GEP(ChatGPT增强后文本)
), ArticleFig(id=1165678287461299098, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=表2, caption=

预训练任务语料组合

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 语料组合
BERT News+Wiki
Gas-kBERT(raw) GAR+GS+GN+
GEP(原始文本)
Gas-kBERTv1.0(GAR+GS+
GN+GEP)
GAR+GS+GN+GEP
(ChatGPT增强后文本)
Gas-kBERTv1.1(+GAR+
GS+GN+GEP)
News+Wiki+GAR+GS+GN+
GEP(ChatGPT增强后文本)
), ArticleFig(id=1165678287507436443, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Table 3, caption=

Fine-tuning task dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
测试任务 数据 训练集 验证集 测试集
命名实
体识别
GS 558,458 260,614 111,692
事故报告 309,782 119,238 2,016
分类 应急预案 1,152 384 128
事故报告 612 116 84
), ArticleFig(id=1165678287666819996, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=表3, caption=

微调任务数据集

, figureFileSmall=null, figureFileBig=null, tableContent=
测试任务 数据 训练集 验证集 测试集
命名实
体识别
GS 558,458 260,614 111,692
事故报告 309,782 119,238 2,016
分类 应急预案 1,152 384 128
事故报告 612 116 84
), ArticleFig(id=1165678287717151645, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Table 4, caption=

NER task test results in the gas pipeline network field %

, figureFileSmall=null, figureFileBig=null, tableContent=
数据 评价
指标
BERT Gas-kBERT(raw) Gas-kBERT v1.0 Gas-kBERT v1.1
(+news_zh_2016+
wiki_zh_2019)
GAR+GS+GN+GEP
(原始文本)
(GAR+GS+GN+GEP)
(ChatGPT增强后文本)
(+news_zh_2016+ wiki_zh_2019+
GAR+GS+GN+GEP)
(ChatGPT增强后文本)
GAR P 11.11 24.22 55.56 55.56
R 13.33 35.19 66.67 33.33
F1 12.12 30.68 60.61 41.67
GN P 30.3 40.02 46.10 40.40
R 16.67 20.15 14.60 43.01
F1 41.67 50.26 69.82 51.34
), ArticleFig(id=1165678287771677598, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=表4, caption=

燃气管网领域NER任务测试结果

, figureFileSmall=null, figureFileBig=null, tableContent=
数据 评价
指标
BERT Gas-kBERT(raw) Gas-kBERT v1.0 Gas-kBERT v1.1
(+news_zh_2016+
wiki_zh_2019)
GAR+GS+GN+GEP
(原始文本)
(GAR+GS+GN+GEP)
(ChatGPT增强后文本)
(+news_zh_2016+ wiki_zh_2019+
GAR+GS+GN+GEP)
(ChatGPT增强后文本)
GAR P 11.11 24.22 55.56 55.56
R 13.33 35.19 66.67 33.33
F1 12.12 30.68 60.61 41.67
GN P 30.3 40.02 46.10 40.40
R 16.67 20.15 14.60 43.01
F1 41.67 50.26 69.82 51.34
), ArticleFig(id=1165678287830397855, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=EN, label=Table 5, caption=

CLASS task results in the field of gas pipeline network %

, figureFileSmall=null, figureFileBig=null, tableContent=
数据 模型 评价
指标
BERT Gas-kBERT(raw) Gas-kBERT v1.0 Gas-kBERT v1.1
类型 (+news_zh_2016+
wiki_zh_2019)
GAR+GS+GN+
GEP(原始文本)
(GAR+GS+GN+GEP)
(ChatGPT增强后文本)
(+news_zh_2016+ wiki_zh_2019+
GAR+GS+GN+GEP)
(ChatGPT增强后文本)
GEP 总则 P 42.42 50.13 62.50 60.87
R 93.33 93.12 100 93.33
F 58.33 70.23 76.92 73.68
应急组
织体系
及职责
P 64.91 40.29 81.82 41.38
R 94.87 90.26 92.31 30.77
F1 77.08 70.33 86.75 35.29
监测、预警 P 100 50.21 66.67 52.63
R 8.33 70.92 83.33 83.33
F1 15.38 60.71 74.07 64.52
应急响应 P 45.95 60.07 63.16 66.67
R 73.91 60.91 52.17 52.17
F1 56.67 56.34 57.14 58.54
信息报告
与发布
P 0 10.19 25 20
R 0 8.28 12.50 12.50
F1 0 10.07 16.67 15.38
善后恢复 P 0 0 0 0
R 0 0 0 0
F1 0 0 0 0
保障措施 P 0 40.43 55.56 47.06
R 0 56.67 71.43 57.14
F1 0 45.13 62.50 51.61
宣传教育、
培训和应
急演练
P 0 0 0 0
R 0 0 0 0
F1 0 0 0 0
附则 P 0 80.27 100 100
R 0 70.19 71.43 71.43
F1 0 79.62 83.33 83.33
GAR 事故伤亡人
员、事故
损失
P 86.21 90.30 92.59 92
R 69.44 87.93 92.59 85.19
F1 76.92 80.51 92.59 88.46
事故原因 P 92.31 93.33 95.65 95.65
R 85.71 95.25 100 100
F1 88.89 96.14 97.78 97.78
性质 P 47.37 60.19 100 63.64
R 100 100 100 100
F1 64.29 70.21 100 77.78
事故发
生经过
P 60.00 70.01 71.43 100
R 54.55 60.06 62.50 62.50
F1 57.14 62.15 66.67 76.92
), ArticleFig(id=1165678287901701024, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1148106728694084402, language=CN, label=表5, caption=

燃气管网领域CLASS任务结果

, figureFileSmall=null, figureFileBig=null, tableContent=
数据 模型 评价
指标
BERT Gas-kBERT(raw) Gas-kBERT v1.0 Gas-kBERT v1.1
类型 (+news_zh_2016+
wiki_zh_2019)
GAR+GS+GN+
GEP(原始文本)
(GAR+GS+GN+GEP)
(ChatGPT增强后文本)
(+news_zh_2016+ wiki_zh_2019+
GAR+GS+GN+GEP)
(ChatGPT增强后文本)
GEP 总则 P 42.42 50.13 62.50 60.87
R 93.33 93.12 100 93.33
F 58.33 70.23 76.92 73.68
应急组
织体系
及职责
P 64.91 40.29 81.82 41.38
R 94.87 90.26 92.31 30.77
F1 77.08 70.33 86.75 35.29
监测、预警 P 100 50.21 66.67 52.63
R 8.33 70.92 83.33 83.33
F1 15.38 60.71 74.07 64.52
应急响应 P 45.95 60.07 63.16 66.67
R 73.91 60.91 52.17 52.17
F1 56.67 56.34 57.14 58.54
信息报告
与发布
P 0 10.19 25 20
R 0 8.28 12.50 12.50
F1 0 10.07 16.67 15.38
善后恢复 P 0 0 0 0
R 0 0 0 0
F1 0 0 0 0
保障措施 P 0 40.43 55.56 47.06
R 0 56.67 71.43 57.14
F1 0 45.13 62.50 51.61
宣传教育、
培训和应
急演练
P 0 0 0 0
R 0 0 0 0
F1 0 0 0 0
附则 P 0 80.27 100 100
R 0 70.19 71.43 71.43
F1 0 79.62 83.33 83.33
GAR 事故伤亡人
员、事故
损失
P 86.21 90.30 92.59 92
R 69.44 87.93 92.59 85.19
F1 76.92 80.51 92.59 88.46
事故原因 P 92.31 93.33 95.65 95.65
R 85.71 95.25 100 100
F1 88.89 96.14 97.78 97.78
性质 P 47.37 60.19 100 63.64
R 100 100 100 100
F1 64.29 70.21 100 77.78
事故发
生经过
P 60.00 70.01 71.43 100
R 54.55 60.06 62.50 62.50
F1 57.14 62.15 66.67 76.92
)], attaches=null, journal=Journal(id=1123942128916217864, delFlag=0, nameCn=中国安全科学学报, nameEn=China Safety Science Journal, nameHistory1=null, nameHistory2=null, issn=1003-3033, eissn=, cn=11-2865/X, coden=null, periodic=0, language=CN, oaType=0, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=fkqsFM6VKlHC4gCtS5XqTw==, journalPrice=null, startedYear=null, abbrevIsoEn=Chin Saf Sci J, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1754269350027, createdBy=null, updatedBy=13701087609, firstLetterCn=C, firstLetterEn=C, subjectCode=Engineering, subjectName=工程, subjectCodeEn=Engineering, subjectNameEn=null, picCn=fkqsFM6VKlHC4gCtS5XqTw==, picEn=SHn9HgqSxtJrOcAxqD++4Q==, jcr=null, cjcr=null, exts=[JournalExt(id=1159052918994595848, language=CN, name=中国安全科学学报, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.cssjj.com.cn/, createdTime=1754269350050, updatedTime=1754269350050, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=http://www.cssjj.com.cn/CN/column/item15.shtml, submissionAuthorUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionReviewUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionCeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionAeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, option={"copyright":""}), JournalExt(id=1159052919040733193, language=EN, name=China Safety Science Journal, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.cssjj.com.cn/EN/1003-3033/home.shtml, createdTime=1754269350061, updatedTime=1754269350061, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=https://synbioj.cip.com.cn/EN/column/column3.shtml, submissionAuthorUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionReviewUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionCeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionAeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, option={"copyright":""})], databaseList=null, tenantJournalId=1146031787341344770, websiteList=[Website(id=1148243202345263519, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146031787341344770, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN, language=CN, createTime=1751692112766, createBy=18614031015, updateTime=1753502583634, updateBy=18614031015, name=《中国安全科学学报》中文站点, tplId=1146099689490845704, title=中国安全科学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148618794941046792, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=articleTextType, value=kx, createTime=1751781661020, updateTime=1751781661020, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794911686661, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=banner, value=null, createTime=1751781661012, updateTime=1751781661012, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794894909444, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=tui0IVO9FMwB61HHtX5scg==, createTime=1751781661008, updateTime=1751781661008, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794932658183, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751781661017, updateTime=1751781661017, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794924269574, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751781661015, updateTime=1751781661015, creator=18614031015, updator=18614031015)]), Website(id=1155836763751993353, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146031787341344770, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN, language=EN, createTime=1753502558893, createBy=18614031015, updateTime=1753524450387, updateBy=18614031015, name=《中国安全科学学报》英文站点, tplId=1146101810881728533, title=China Safety Science Journal, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155895925743669425, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=articleTextType, value=kx, createTime=1753516664205, updateTime=1753516664205, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925722697902, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=banner, value=null, createTime=1753516664200, updateTime=1753516664200, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925714309293, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=tui0IVO9FMwB61HHtX5scg==, createTime=1753516664198, updateTime=1753516664198, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925735280816, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753516664203, updateTime=1753516664203, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925731086511, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753516664202, updateTime=1753516664202, creator=18614031015, updator=18614031015)])], journalTitle=中国安全科学学报, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=China Safety Science Journal, journalPhotoCn=fkqsFM6VKlHC4gCtS5XqTw==, journalPhotoEn=SHn9HgqSxtJrOcAxqD++4Q==, journalFirstLetter=C, journalRecommend=null, journalNew=null, journalCollection=1, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=null, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/10.16265/j.cnki.issn1003-3033.2025.03.0223, detailUrlEn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN/10.16265/j.cnki.issn1003-3033.2025.03.0223, pdfUrlCn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/PDF/10.16265/j.cnki.issn1003-3033.2025.03.0223, pdfUrlEn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN/PDF/10.16265/j.cnki.issn1003-3033.2025.03.0223, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于知识注入的燃气知识双向变换器模型
收藏切换
PDF下载
柳晓昱 , 庄育锋 ** , 赵兴昊 , 王珂璠 , 张国开
中国安全科学学报 | 公共安全 2025,35(3): 204-211
收起
收藏切换
中国安全科学学报 | 公共安全 2025, 35(3): 204-211
基于知识注入的燃气知识双向变换器模型
全屏
柳晓昱 , 庄育锋** , 赵兴昊, 王珂璠, 张国开
作者信息
  • 北京邮电大学 智能工程与自动化学院,北京 100876
  • 柳晓昱 (2000—),女,天津人,硕士研究生,主要研究方向为文本挖掘与物流信息技术及工程应用。E-mail:

通讯作者:

** 庄育锋(1972—),男,上海人,博士,教授,主要从事智能控制与装备安全等方面的研究。E-mail:
Gas knowledge bidirectional encoder representations from transformers model based on knowledge injection
Xiaoyu LIU , Yufeng ZHUANG** , Xinghao ZHAO, Kefan WANG, Guokai ZHANG
Affiliations
  • School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China
出版时间: 2025-03-28 doi: 10.16265/j.cnki.issn1003-3033.2025.03.0223
文章导航
收藏切换

为提高燃气管网领域的应急管理水平,提出燃气知识双向变换器(Gas-kBERT)模型。该模型结合聊天生成预训练转换器(ChatGPT)扩充的燃气管网领域数据,以及构建的中文燃气语言理解-三元组(CGLU-Spo)和相关语料库,通过改变模型的掩码(MASK)机制,成功将领域知识注入模型中。考虑到燃气管网领域的专业性和特殊性,Gas-kBERT在不同规模和内容的语料库上进行预训练,并在燃气管网领域的命名实体识别和分类任务上进行微调。结果表明:与通用的双向变换器(BERT)模型相比,Gas-kBERT在燃气管网领域的文本挖掘任务中F1值表现出显著的提升。在命名实体识别任务中,F1值提高29.55%;在文本分类任务中,F1值提升高达83.33%。由此证明Gas-kBERT模型在燃气管网领域的文本挖掘任务中具有出色的表现。

燃气管网  /  燃气知识双向变换器(Gas-kBERT)模型  /  自然语言处理(NLP)  /  知识注入  /  双向变换器(BERT)模型

In order to enhance emergency management in the field of gas pipeline networks,Gas-kBERT model was proposed. The model incorporated data from the gas pipeline network field expanded by Chat Generative Pre-Trained Transformer,(ChatGPT)and Chinese Gas Language Understanding Subject-Predicate-Object(CGLU-Spo) and related corpora were constructed in this field. By altering the model's masking (MASK) mechanism,domain knowledge was successfully injected into the model. Considering the professionalism and specificity of the gas pipeline network field,Gas-kBERT was pre-trained on various scales and contents of corpora and fine-tuned on named entity recognition and classification tasks within this field. Experimental results demonstrated that,compared to the general BERT model,Gas-kBERT exhibited significant performance improvements in F1-score in text mining tasks in the gas pipeline network field. Specifically,in the named entity recognition task,the F1-score was increased by 29.55%,and in the text classification task,the F1-score improvement reached up to 83.33%. This study proves that the Gas-kBERT model performs exceptionally well in text mining tasks in the gas pipeline network field.

gas pipeline networks  /  gas knowledge bidirectional encoder representations from transformers(Gas-kBERT)model  /  natural language processing(NLP)  /  knowledge injection  /  bidirectional encoder representations from transformers (BERT)
柳晓昱, 庄育锋, 赵兴昊, 王珂璠, 张国开. 基于知识注入的燃气知识双向变换器模型. 中国安全科学学报, 2025 , 35 (3) : 204 -211 . DOI: 10.16265/j.cnki.issn1003-3033.2025.03.0223
Xiaoyu LIU, Yufeng ZHUANG, Xinghao ZHAO, Kefan WANG, Guokai ZHANG. Gas knowledge bidirectional encoder representations from transformers model based on knowledge injection[J]. China Safety Science Journal, 2025 , 35 (3) : 204 -211 . DOI: 10.16265/j.cnki.issn1003-3033.2025.03.0223
燃气作为高效、清洁的能源,在工业化国家中占据重要地位。随着“西气东输”一线工程的成功投产以及川气东输、西气东输等工程的持续推进,我国已构建了庞大且复杂的燃气输配管道系统。但由于存在施工质量、管道使用环境与管道老化等多种因素,近年来燃气管网事故频发,对城市公共安全造成极大威胁,并造成重大经济损失[1]。针对这一现状,“十四五”规划明确指出,未来10~20年,我国管道储运将处于稳定增长期,其中,管网建设将是发展重点。同时,也提出加强管网安全管理和风险防范的迫切需求。
信息技术的发展为燃气管网安全管理提供新的方法,如风险评价[2]、可视化技术[3]、3D视觉技术[4]等。然而,燃气管网领域的文本数据,如记录、安全检查和经验教训等,尚未得到充分利用。这些非结构化的信息采用手动分析文本非常耗时[5],需要通过自动化手段进行整理和筛选。自然语言处理(Natural Language Processing,NLP)技术的兴起,为其提供了新的解决方案[6],通过迁移学习方法从多样化数据源中提取有价值信息[7-9]。随着NLP的发展,预训练语言模型经历了从早期word embedding[10]到现代上下文感知预训练模型的演变[11-12]。其中,双向变换器(Bidirectional Encoder Representations from Transformers,BERT)模型以其掩码语言模型和下一句预测任务展示了强大的泛化能力。但当涉及专业术语和特定词汇时,通用预训练语言模型就难以准确理解特定专业领域内的文本[13]。许多研究尝试将外部知识融入以BERT为代表的预训练语言模型中[14-17],但目前在燃气管网领域缺乏针对该领域的定制化预训练模型。考虑到燃气管网领域的专业性和复杂性,定制化预训练模型在事故分析、风险评估和管理决策等方面具有巨大应用潜力,有望成为管理人员处理大量文档和案例的智能工具,以提高工作效率和决策质量。
鉴于此,笔者提出燃气知识BERT(Gas-knowlodge,Gas-kBERT)模型,以解决通用模型对特殊术语理解困难的问题,为验证Gas-kBERT模型的有效性,进行命名实体识别(Named Entity Recognition,NER)和文本分类(Classification,CLASS)2个任务的试验,以期能够在相关领域实现技术创新和成果突破。
Gas-kBERT是一种自我监督的预训练模型,旨在更好地学习燃气管网领域知识,但是不同于BERT的输入策略。首先,使用聊天生成预训练转换器(Chat Generative Pre-Trained Transformer,ChatGPT)工具生成额外文本数据,以提升模型训练效果。其次,基于大量燃气管网相关语料生成三元组,进一步丰富文本内容[18]。然后,使用2种不同的掩码机制将燃气管网领域知识注入模型中。最后,将输入文档分割成片段作为正样本,并从其他样本中随机抽取句子作为负样本,用于下一句预测。模型结构如图1所示。
ChatGPT经过大规模数据训练,并通过大量人类反馈进行强化训练,使模型具有人类语言的自然性。对NLP模型的泛化性能、对抗攻击、干扰波动有很好的提升作用。
Gas-kBERT首先利用ChatGPT工具将训练样本中的每个句子重述为多个语义相近但概念不同的样本,增强后的文本可有效用于下游模型的训练,显著提高模型的泛化能力和准确性,为后续的任务提供更好的支持。其次,通过构建燃气管网领域三元组图谱—中文燃气语言理解-三元组(Chinese Gas Language Understanding Subject-Predicate-Object,CGLU-Spo)扩充领域专业术语和行业词汇。通过mask技术掩盖三元组中的实体词,并通过注入三元组中的关系属性知识来深入学习这些实体词的含义,从而增强模型对燃气管网领域的理解能力,这一过程如图2所示。最后CGLU-Spo图谱基于现有燃气管网知识库,通过工具标注三元组信息,获得3 990条三元组,这些三元组结合预训练使用的全部数据,构建一个覆盖面广泛的三元组语料。
该模型主要采用WWM和WEM等2种掩码机制(图3),这2种方法有效增强了模型对输入文本的理解能力。
1) WWM。此机制不再仅遮蔽单个标记,而是选择遮蔽整个单词。这种策略在处理依赖单词上下文信息的任务时尤为有效,相比于单个标记的遮蔽,WWM能够保留单词的完整语义和上下文信息,从而显著提升模型对文本内容的理解能力,并在相关任务中提升模型性能。
2) WEM。与BERT遮蔽机制不同,此机制是一种确定性的策略,利用燃气管网领域特定实体的知识来增强模型。具体而言,利用预先构建的燃气管网领域三元组语料库,其中,包含关键实体,如“燃气引入管”等;模型在处理文本时,会识别句子中是否包含这些三元组中的实体,并进行遮蔽操作。通过这种方式能够有效地将领域相关的实体和专业知识注入模型中,大幅提升模型在燃气管网相关任务中的表现。
通过引入上述的掩码机制,模型可以学习如何使用属性和关系来识别和理解实体,并将这些知识融合到模型中。
BERT作为通用语言模型,主要在通用领域语料库上进行预训练,但由于缺乏直接支持燃气管网领域的数据库,大量燃气管网领域数据未得到充分利用,使得在燃气管网文本挖掘中因领域差异表现受限。为此,基于BERT模型使用燃气管网领域数据预训练模型并做微调,预训练文本见表1,整体训练流程如图4所示。
通过最小的架构修改,Gas-kBERT能够胜任多种燃气管网文本挖掘任务。并在NER和CLASS这2个关键任务中进行微调,评估指标采用精确率P、召回率RF值。
1) 试验数据集获取与处理。通过爬虫等工具收集燃气管网领域的多样化数据,包括事故报告、标准以及事故新闻等(表2)。为保证模型训练效果,数据集经过预处理,包括去除重复、无效和空文本、分词、标点符号和停用词[19]。最终获得的预训练和微调任务文本语料见表2表3
2) 三元组语料CGLU-Spo的部署。为提取结构化知识,基于现有知识库并选取涉及燃气管网文本的网站,爬取相关词条。借助标注工具,构建3 990条燃气管网领域的三元组语料。
3) 模型参数设置。为与BERT进行比较,采用相同的模型设置,模型包含12层,12个self-attention heads,768维的hidden size,1 000 000个step,每个batch包含256条序列。learning rate为1×10-4β1 = 0.9、β2 = 0.999。learning rate前10 000个steps使用rate warmup,之后开始线性衰减learning rate。对所有layer使用0.1概率的dropout。使用的激活函数为gelu。
4) 试验环境。试验采用的图形处理器为NVIDIATI Tesla T4,代码通过pycharm平台进行编程,开发语言Python,深度学习框架为Tensorflow1.15.0,torch1.10.2。
5) 对比模型。为验证Gas-kBERT模型在下游任务的优越性,引入对照组BERT模型,同时为验证预训练数据内容对Gas-kBERT模型的影响,引入Gas-kBERT(raw)、Gas-kBERTv1.0和Gas-kBERTv1.1作为对照组,模型均在相同的试验环境下以相同的方式进行训练。
燃气管网相关的报告具有很强的非结构化性,通过NER技术,可以从这些报告中提取关键信息,如事故发生时间、地点、原因以及性质等。通过对比在不同数据集上进行预训练的模型在NER任务上的表现,得出试验结果,具体见表4。将结论分为4类:①Gas-kBERT(包括Gas-kBERTv1.0和Gas-kBERTv1.1)和Gas-kBERT(raw)在所有数据集上的F分数均优于BERT,这表明使用燃气管网领域语料进行预训练的模型,在燃气管网领域的文本挖掘任务中表现更为出色。②在GAR数据测试任务中Gas-kBERTv1.0取得更高的F值,相较于Gas-kBERTv1.1和BERT上升18.94%和48.49%。在GN数据测试任务中分别提升18.48%和28.15%。这一结果说明,预训练的数据并不是越多越好,而是越有针对性越好。③Gas-kBERT(raw)在所有数据集上的NER任务测试结果的F分数均优于Gas-kBERT,这表明:通过使用ChatGPT增强文本能够提升模型的训练效果。④几个模型在数据集上的得分普遍相对较低,原因可能包括:标注数据集有限;模型参数尚未优化调整以达到最佳性能;标注数据集可能存在错误、标签不一致或模糊,这可能影响模型的训练和评估质量。
分类任务是根据燃气管网相关文本对内容类别进行分类。在这一阶段,采用相同的数据,结果见表5。从中可以将结果分为3类:①Gas-kBERT在多数与燃气管网CLASS任务相关的数据集上较BERT具有更高的F1分数,可以看出,用于燃气管网语料进行预训练的模型,在燃气管网领域文本挖掘任务中通常表现更好。然而,也注意到在应急组织体系及职责和应急响应2个具体任务上,BERT的表现优于Gas-kBERT(raw),但Gas-kBERT表现又优于BERT,这表明:在模型训练过程中,仅注入相关领域文本是不够的,更重要的是要让模型能够有效地学习和理解这些知识。②Gas-kBERT较Gas-kBERT(raw)具有更高的F1分数,再次证明使用ChatgGPT增强文本是有效的。③Gas-kBERT v1.1在所有燃气管网CLASS数据集上的训练得分相对于Gas-kBERT v1.0较低,可能原因为:预训练数据可能过于多样化和杂乱,包括通用和燃气管网领域数据;在不同数据集上可能需要调整不同的训练参数,以获得最佳模型性能。
1) Gas-kBERT模型解决了传统通用模型无法充分识别专业领域文本的问题。试验结果表明:Gas-kBERT模型在燃气管网文本挖掘任务中明显优于BERT和其他基准模型的效果,尤其是在使用包含(GAR+GS+GN+GEP)数据的Gas-kBERT v1.0模型中,F1分数提升达到83.33%。
2) 从知识注入角度思考燃气领域预训练语言模型,通过改变MASK策略、注入相关领域三元组、结合ChatGPT有效提升了模型在燃气领域任务中的精确率P、召回率RF1分数。
3) 文中还存在一些限制,如微调任务相对较少、知识图谱的引入方式简单等。为进一步提升模型的性能和泛化能力,未来的工作中将扩充测试任务数据集、考虑注入的知识在文本中的位置以及探索更多技术手段来优化模型表现。
  • 国家自然科学基金资助(52478123)
参考文献 引证文献
排序方式:
[1]
MONTIEL H, VÍLCHEZ JA, ARNALDOS J, et al. Historical analysis of accidents in the transportation of natural gas[J]. Journal of Hazardous Materials, 1996, 51 (1/3):77-92.
[2]
郭秀军, 刘嘉柠. 智能化燃气管网监测与管理[J]. 网络安全和信息化, 2024(1):70-72.
[3]
李兰. 基于视觉检测与图像处理分析技术的燃气管道检测方法研究[D]. 南京: 东南大学, 2019.
LI Lan. Study on the gas pipeline detection method based on visual inspection and image processing analysis technology[D]. Nanjing: Southeast University, 2019.
[4]
苟爽. 基于机器视觉的燃气管道内环境感知技术研究[D]. 昆明: 昆明理工大学, 2018.
GOU Shuang. Research on internal environmental perception technology of gas pipeline based on machine vision[D]. Kunming: Kunming University of Science and Technology, 2018.
[5]
周吉祥, 张浩. 智能管道建设下的城镇燃气管道完整性管理[J]. 煤气与热力, 2021, 41(2):32-36,46.
ZHOU Jixiang, ZHANG Hao. Integrity management of urban gas pipeline under construction of intelligent pipeline[J]. Gas & Heat, 2021, 41(2):32-36,46.
[6]
李思洁, 王亚慧, 张子豪. 燃气输配突发事件应急处置的知识图谱构建[J]. 消防科学与技术, 2022, 41(6):812-817.
LI Sijie, WANG Yahui, ZHANG Zihao. Construction of knowledge graph for emergency disposal of gas transmission and distribution emergencies[J]. Fire Science and Technology, 2022, 41(6):812-817.
[7]
TIWARY U S, SIDDIQUI T. Natural language processing and information retrieval[M]. New York: Oxford University Press, 2008:656.
[8]
WANG Chenglong, JIANG Feijun, YANG Hongxia. A hybrid framework for text modeling with convolutional RNN[C]. The 23rd ACM SIGKDD International Conference, 2017: DOI:10.1145/3097983.3098140.
[9]
QIN Ya, SHEN Guowei, ZHAO Wenbo, et al. A network security entity recognition method based on feature template and CNN-BiLSTM-recognition method based on feature template and CNN-BiLSTM-CRF[ J]. Frontiers of Information Technology and Electronic Engi-neering, 2019, 20(6):13-19.
[10]
RAO N V, KRISHNA B T. Automatic modulation recognition using machine learning techniques: a review[C]// KALYA S, KULKARNI M, SHIVAPRAKASHA K S. Advances in VLSI, Signal Processing, Power Electronics, IoT, Communication and Embedded Systems. Singapore: Springer Singapore, 2021: 145-154.
[11]
MA Zhiliang, CAI Shiyao, MAO Na, et al. Construction quality management based on a collaborative system using BIM and indoor positioning[J]. Automation in Construction, 2018, 92:35-45.
[12]
PETERS M, NEUMANN M, IYYER M, et al. Deep contextualized word representations[C]. The 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,2018:2227-2237.
[13]
DEVLIN J, CHANG Mingwei, LEE Kenton, et al. BERT: pre-training of deep bidirectional transformers for language understanding[C]. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019.DOI:10.48550/arXiv.1810.04805.
[14]
邓存彬, 虞慧群, 范贵生, 等. 燃气客服热线的中文文本情感分析[J]. 华东理工大学学报:自然科学版, 2019, 45(1):140-147.
DENG Cunbin, YU Huiqun, FAN Guisheng, et al. Sentiment analysis of Chinese texts for gas customer service hotline[J]. Journal of East China University of Science and Technology:Natural Science Edition, 2019, 45(1):140-147.
[15]
王明达, 张榜, 吴志生, 等. 基于强化学习的城镇燃气事故信息抽取方法[J]. 中国安全生产科学技术, 2023, 19(3):39-45.
WANG Mingda, ZHANG Bang, WU Zhisheng, et al. Information extraction method of urban gas accidents based on reinforcement learning[J]. Journal of Safety Science and Technology, 2023, 19(3):39-45.
[16]
LEE J H, YOON W J, KIM S D, et al. BioBERT:a pre-trained biomedical language representation model for biomedical text mining[J]. Bioinformatics, 2020, 36(4):1234-1240.
[17]
WANG Xuan, ZHANG Xiang, REN Yuhao, et al. Cross-type biomedical named entity recognition with deep multi-task learning[J]. Bioinformatics, 2019, 35(10):1745-1752.
[18]
宋敦江, 杨霖, 钟少波. 基于 BERT 的灾害三元组信息抽取优化研究[J]. 中国安全科学学报, 2022, 32(2) : 115-120.
SONG Dunjiang, YANG Lin, ZHONG Shaobo. Research on optimization of disaster triplet information extraction based on BERT[J]. China Safety Science Journal, 2022, 32(2): 115-120.
[19]
郑彬彬, 冯婷婷, 王佳贺, 等. 基于文本挖掘的城镇燃气事故致因及关联分析[J]. 中国安全科学学报, 2023, 33(7):190-195.
ZHENG Binbin, FENG Tingting, WANG Jiahe, et al. Causes and correlation analysis of urban gas accidents based on textmining[J]. China Safety Science Journal, 2023, 33(7): 190-195.
2025年第35卷第3期
PDF下载
333
135
引用本文
BibTeX
文章信息
doi: 10.16265/j.cnki.issn1003-3033.2025.03.0223
  • 接收时间:2024-10-14
  • 首发时间:2025-07-05
  • 出版时间:2025-03-28
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-10-14
  • 修回日期:2024-12-18
基金
国家自然科学基金资助(52478123)
作者信息
    北京邮电大学 智能工程与自动化学院,北京 100876

通讯作者:

** 庄育锋(1972—),男,上海人,博士,教授,主要从事智能控制与装备安全等方面的研究。E-mail:
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/10.16265/j.cnki.issn1003-3033.2025.03.0223
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏