From the view of language itself rather than the factors that affect language,the author illuminates how linguistic context plays its role in translation.In the process of translation,words and phrases may have more t...From the view of language itself rather than the factors that affect language,the author illuminates how linguistic context plays its role in translation.In the process of translation,words and phrases may have more than one meaning,which need translators to fix the appropriate one the discourse demands.Without sentence group,sentences can be ambiguous too.Paragraphs and texts have structures which words and sentences do not.Translators need to choose proper words to express the meaning properly.展开更多
In this paper,we'll explore the role of context in the interpretation of a piece of discourse. We'll present three context models put forward respectively by Firth,Hymes and Lewis. We'll also take into con...In this paper,we'll explore the role of context in the interpretation of a piece of discourse. We'll present three context models put forward respectively by Firth,Hymes and Lewis. We'll also take into consideration the features of cotext and expanding context. At last,we'll throw some light on the problem of what aspects of the apparently illimitable features of context are relevant to a particular communicative act.展开更多
大语言模型(large language models,LLMs)在知识存储方面不断增强的能力展示了其作为知识库的潜在效用.然而,任何给定的提示只能提供大语言模型所涵盖知识的下限估计.在语言模型即知识库(language models as knowledge bases,LMs-as-KBs...大语言模型(large language models,LLMs)在知识存储方面不断增强的能力展示了其作为知识库的潜在效用.然而,任何给定的提示只能提供大语言模型所涵盖知识的下限估计.在语言模型即知识库(language models as knowledge bases,LMs-as-KBs)的场景中,先前的提示学习方法忽略了查询风格对模型表现的影响.揭示了大语言模型确实具有与查询风格相关的可学习偏好,并且利用大语言模型的这种特性引入了查询风格自适应转换(adaptive query style transfer,ARES)方法,通过适应大语言模型的偏好来增强其知识查询的表现. ARES方法从构造查询候选集开始,通过改写实现多种表达风格的纳入.随后,ARES训练一个评估器来学习并识别大语言模型对查询风格的偏好,评估查询候选集并选择潜在最优查询.在多个数据集上进行的实验表明了该方法在提高大语言模型即知识库服务上查询准确率的有效性,增量对比原始模型与3个基线方法分别实现了平均2.26个百分点,1.68个百分点,1.19个百分点,1.17个百分点的提升,这表明ARES可以与其他方法有效地结合使用,从而实现多角度的查询表现增强.展开更多
语言学界对连词“and”关联构成假拟祈使句现象(如“Come closer and I'll shoot.”)争议颇大,然而鲜有研究采用真实语料、跨越句子对该现象进行探讨。本文基于系统功能语言学的语境观、功能观和系统论,选取enTenTen18语料库中动词...语言学界对连词“and”关联构成假拟祈使句现象(如“Come closer and I'll shoot.”)争议颇大,然而鲜有研究采用真实语料、跨越句子对该现象进行探讨。本文基于系统功能语言学的语境观、功能观和系统论,选取enTenTen18语料库中动词短语-and-陈述句联为语料,从语篇语义视角出发,探讨假拟祈使句在不同语篇语境下的语义解读及其语篇合谋功能。研究发现:情景语境和上下文语境为假拟祈使句的语义分化提供了资源配置;在语篇元功能的促成作用及其与人际元功能的合力作用下,说话人受语篇意图和语境变化的驱动选择不同的假拟祈使句来组构语篇;假拟祈使句在自然语篇中具有获得首位突显、与if条件句等形式同义置换、引出新话题的作用,能为人际语义潜势多样化体现提供语法资源。展开更多
针对复杂环境下含标签货物实时记录困难的问题,提出一种面向视觉物联网(visual Internet of Things,VIoT)的文本检测方法。在视觉物联网中设计并引入基于全局上下文注意力和坐标注意力的文本检测网络(text detection network based on g...针对复杂环境下含标签货物实时记录困难的问题,提出一种面向视觉物联网(visual Internet of Things,VIoT)的文本检测方法。在视觉物联网中设计并引入基于全局上下文注意力和坐标注意力的文本检测网络(text detection network based on global context attention and coordinate attention,GCANet),首先提出一种改进型坐标注意力模块,通过水平和垂直2个并行的一维池化操作,避免了因二维全局池化造成的位置信息丢失;然后引入全局上下文注意力模块,避免在复杂的背景对文本检测的影响,并防止密集或较远间隔的文本被错误地检测。该系统中提出的GCANet在公共数据集ICDAR2015、MSRA-TD500和Total-Text上的综合指标F值分别达到87.4%、86.9%和86.3%。在工业标签数据集Label-Text上平均准确率、平均召回率和平均F值分别达到93.4%、90.9%和92.1%。此外,GCANet在矿井下的标签数据集Mine-Text上准确率、召回率和F值分别达到94.4%、84.9%和89.9%。实验结果表明,本文提出的面向视觉物联网的文本检测方法效果优异。展开更多
文摘From the view of language itself rather than the factors that affect language,the author illuminates how linguistic context plays its role in translation.In the process of translation,words and phrases may have more than one meaning,which need translators to fix the appropriate one the discourse demands.Without sentence group,sentences can be ambiguous too.Paragraphs and texts have structures which words and sentences do not.Translators need to choose proper words to express the meaning properly.
文摘In this paper,we'll explore the role of context in the interpretation of a piece of discourse. We'll present three context models put forward respectively by Firth,Hymes and Lewis. We'll also take into consideration the features of cotext and expanding context. At last,we'll throw some light on the problem of what aspects of the apparently illimitable features of context are relevant to a particular communicative act.
文摘语言学界对连词“and”关联构成假拟祈使句现象(如“Come closer and I'll shoot.”)争议颇大,然而鲜有研究采用真实语料、跨越句子对该现象进行探讨。本文基于系统功能语言学的语境观、功能观和系统论,选取enTenTen18语料库中动词短语-and-陈述句联为语料,从语篇语义视角出发,探讨假拟祈使句在不同语篇语境下的语义解读及其语篇合谋功能。研究发现:情景语境和上下文语境为假拟祈使句的语义分化提供了资源配置;在语篇元功能的促成作用及其与人际元功能的合力作用下,说话人受语篇意图和语境变化的驱动选择不同的假拟祈使句来组构语篇;假拟祈使句在自然语篇中具有获得首位突显、与if条件句等形式同义置换、引出新话题的作用,能为人际语义潜势多样化体现提供语法资源。
文摘针对复杂环境下含标签货物实时记录困难的问题,提出一种面向视觉物联网(visual Internet of Things,VIoT)的文本检测方法。在视觉物联网中设计并引入基于全局上下文注意力和坐标注意力的文本检测网络(text detection network based on global context attention and coordinate attention,GCANet),首先提出一种改进型坐标注意力模块,通过水平和垂直2个并行的一维池化操作,避免了因二维全局池化造成的位置信息丢失;然后引入全局上下文注意力模块,避免在复杂的背景对文本检测的影响,并防止密集或较远间隔的文本被错误地检测。该系统中提出的GCANet在公共数据集ICDAR2015、MSRA-TD500和Total-Text上的综合指标F值分别达到87.4%、86.9%和86.3%。在工业标签数据集Label-Text上平均准确率、平均召回率和平均F值分别达到93.4%、90.9%和92.1%。此外,GCANet在矿井下的标签数据集Mine-Text上准确率、召回率和F值分别达到94.4%、84.9%和89.9%。实验结果表明,本文提出的面向视觉物联网的文本检测方法效果优异。
