As H1N1 influenza rages across the world from the very beginning of mid 2009,lots of news reports focus on it.This essay mainly studies the stylistics features of News English by reading news reports on H1N1.The study...As H1N1 influenza rages across the world from the very beginning of mid 2009,lots of news reports focus on it.This essay mainly studies the stylistics features of News English by reading news reports on H1N1.The study tries to find lexical features,sentence features and grammatical features of News English.Through the study we find that News English has its own stylistics features in words,sentences and grammar.展开更多
现有的医学健康问句数据大多数都是短文本,但短文本存在特征稀疏的局限性。对此,提出一种融合特征的方法,首先通过基于变换器的双向编码器表征技术(Bidirectional Encoder Representations from Transformers,BERT)字符级特征的输出取...现有的医学健康问句数据大多数都是短文本,但短文本存在特征稀疏的局限性。对此,提出一种融合特征的方法,首先通过基于变换器的双向编码器表征技术(Bidirectional Encoder Representations from Transformers,BERT)字符级特征的输出取平均并与BERT句子级特征的输出进行拼接,然后使用分类器进行分类。实验结果表明,本模型可以有效地提高模型提取特征的能力,在处理Kesci公众健康问句分类数据集上F1值达到83.92%,在处理中文健康公众问句数据集时F1值达到87%。展开更多
文摘As H1N1 influenza rages across the world from the very beginning of mid 2009,lots of news reports focus on it.This essay mainly studies the stylistics features of News English by reading news reports on H1N1.The study tries to find lexical features,sentence features and grammatical features of News English.Through the study we find that News English has its own stylistics features in words,sentences and grammar.
文摘现有的医学健康问句数据大多数都是短文本,但短文本存在特征稀疏的局限性。对此,提出一种融合特征的方法,首先通过基于变换器的双向编码器表征技术(Bidirectional Encoder Representations from Transformers,BERT)字符级特征的输出取平均并与BERT句子级特征的输出进行拼接,然后使用分类器进行分类。实验结果表明,本模型可以有效地提高模型提取特征的能力,在处理Kesci公众健康问句分类数据集上F1值达到83.92%,在处理中文健康公众问句数据集时F1值达到87%。
