The year 2019 marks the seventh anniversary of Shinzo Abe's return to office,alongside the administration's obsession with Constitutional amendment.During the 25th House of Councillors selection in July,Prime ...The year 2019 marks the seventh anniversary of Shinzo Abe's return to office,alongside the administration's obsession with Constitutional amendment.During the 25th House of Councillors selection in July,Prime Minister Abe changed the rules applied for the four national elections in 2012,wrote the Self-Defense Forces into the Constitution and the Liberal Democratic Party(LDP)convention,and deliberately made the amendment a focal point of controversy in his election campaign.However,when the results of the election came out,the pro-amendment forces,with the LDP at their core,didn't reach the two-thirds majority threshold required for Constitutional revision.Going forwards,the Abe-led Liberal Democratic Party will strive to consolidate Constitutional amendment forces in the National Diet,hoping to cross the majority threshold,and at the same time to coerce opposition parties to comply with passing LDP draft revisions to the Constitution in the National Diet as soon as possible,in order to submit these to national referendum.In the meantime,preparatory works for this referendum are being carried out through the use of political resources to mobilize public support.Whether Abe can reach the political objective of Constitutional amendment during his current term of office and how the Constitutional issue in Japan will play out is sure to attract global attention.展开更多
电网现场作业风险等级影响着作业的安全性和经济性,目前风险等级评级主要依赖于人工评级,存在一定的错误率。以国网公司的作业风险分级表与实际现场历史文本库作为研究对象,在充分分析各自文本特点的基础上,提出了基于BERT(bidirectiona...电网现场作业风险等级影响着作业的安全性和经济性,目前风险等级评级主要依赖于人工评级,存在一定的错误率。以国网公司的作业风险分级表与实际现场历史文本库作为研究对象,在充分分析各自文本特点的基础上,提出了基于BERT(bidirectional encoder representations from transformers)模型的电网现场作业风险自动评级方法。考虑作业风险分级表的概括性,提出对作业风险分级表进行文本增强;考虑实际现场历史文本库中存在一定比例的冗余内容和错误评级,提出冗余文本去除方法和基于文本挖掘技术的风险等级纠错方法;以经过上述方法处理后的2类现场作业文本为样本,构建了基于BERT单文本主题分类微调模型的文本分类模型,实现了对电网现场作业风险的自动评级。算例对比了该文模型与其他分类模型在不同样本集下的评级效果,表明BERT模型能够更全面地获取作业文本的语义特征,分级表的文本增强方法和历史库的风险等级纠错方法能提升评级效果,而冗余文本去除会导致作业文本样本数量减少较多,反而降低评级效果。上述结果表明,该文方法能够实现电网现场作业文本的自动评级,并具有较高的准确性。展开更多
基金This paper presents the initial findings of a major research project entitled"Research on Postwar Japanese Politics,Diplomacy and Future Trends/5 funded by the Ministry of Education(Project Number:14JZD033).
文摘The year 2019 marks the seventh anniversary of Shinzo Abe's return to office,alongside the administration's obsession with Constitutional amendment.During the 25th House of Councillors selection in July,Prime Minister Abe changed the rules applied for the four national elections in 2012,wrote the Self-Defense Forces into the Constitution and the Liberal Democratic Party(LDP)convention,and deliberately made the amendment a focal point of controversy in his election campaign.However,when the results of the election came out,the pro-amendment forces,with the LDP at their core,didn't reach the two-thirds majority threshold required for Constitutional revision.Going forwards,the Abe-led Liberal Democratic Party will strive to consolidate Constitutional amendment forces in the National Diet,hoping to cross the majority threshold,and at the same time to coerce opposition parties to comply with passing LDP draft revisions to the Constitution in the National Diet as soon as possible,in order to submit these to national referendum.In the meantime,preparatory works for this referendum are being carried out through the use of political resources to mobilize public support.Whether Abe can reach the political objective of Constitutional amendment during his current term of office and how the Constitutional issue in Japan will play out is sure to attract global attention.
文摘电网现场作业风险等级影响着作业的安全性和经济性,目前风险等级评级主要依赖于人工评级,存在一定的错误率。以国网公司的作业风险分级表与实际现场历史文本库作为研究对象,在充分分析各自文本特点的基础上,提出了基于BERT(bidirectional encoder representations from transformers)模型的电网现场作业风险自动评级方法。考虑作业风险分级表的概括性,提出对作业风险分级表进行文本增强;考虑实际现场历史文本库中存在一定比例的冗余内容和错误评级,提出冗余文本去除方法和基于文本挖掘技术的风险等级纠错方法;以经过上述方法处理后的2类现场作业文本为样本,构建了基于BERT单文本主题分类微调模型的文本分类模型,实现了对电网现场作业风险的自动评级。算例对比了该文模型与其他分类模型在不同样本集下的评级效果,表明BERT模型能够更全面地获取作业文本的语义特征,分级表的文本增强方法和历史库的风险等级纠错方法能提升评级效果,而冗余文本去除会导致作业文本样本数量减少较多,反而降低评级效果。上述结果表明,该文方法能够实现电网现场作业文本的自动评级,并具有较高的准确性。
