“大型藏文基础语料库”数据分析1

语料库是存储于计算机中并可利用计算机进行检索、查询、分析的语言素材总体,是理想的语言知识资源。国家非常重视语料库的建设,包括少数民族语种语料库。“大型藏文基础语料库建设”是教育部、国家语委一个大型的藏语通用语料库的建设项目,本文结合该语料库的建设分析了语料库的样本类别、语料样本数据的时间分布、领域分布、来源及地域分布,也说明了语料库词性标注的标记集和标注样本的领域分布,为使用该语料库的人员提供了一个参考数据。     

基于注意力机制的包装命名实体识别

目的 为了解决包装行业相关文本命名实体识别困难问题,提出在BiLSTM（Bidirectional Long Short-Term Memory）神经网络中加入注意力机制（Attention）和字词联合特征,构建一种基于注意力机制的BiLSTM深度学习模型（简称Attention-BiLSTM）,以识别包装命名实体。方法 首先构建包装领域词典匹配包装语料中词语的类别特征,同时将包装语料转换为字特征和词特征联合的向量特征,并且在过程中加入POS（词性）信息。然后将以上特征联合馈送到BiLSTM网络,以获取文本的全局特征,并利用注意力机制获取局部特征。最后根据文本的全局特征和局部特征使用CRF（Conditional Random Field）解码整个句子的最优标注序列。结果 通过对《中国包装网》新闻数据集的实验,获得了85.6%的F值。结论 所提方法在包装命名实体识别中优于传统方法。     

面向石油化工领域的语料库建设研究

基于专家内省方法获取领域知识已无法满足要求,试图发挥内省和统计两种方法的优势,提出包含生语料级、词汇级、句子级和篇章级4个层级的领域语料库设计框架,使每一个层级的语料可以独立地进行NLP分析和服务高一个层级语料.在此设计框架下,建立了大规模的石油化工领域语料库,为石油化工行业的知识获取和分析提供基础资源和素材,支撑石化领域知识工程项目的研究与应用.     

应用粗糙集理论提取特征的词性标注模型

针对词性标注中的复杂特征提取问题,应用粗糙集理论(rough sets),有效地挖掘了包括长距离特征在内的复杂特征,并有效地处理了语料库噪声问题.最后,将这些特征融合于最大熵模型中,训练时按模型整体性能为其分配权重.开放实验表明:增加粗规则后获得96.29%的标注精度,相比原有模型提高了0.83%.     

基于序列标注的中医症状名识别技术研究

传统中医学博大精深,随着计算机技术在医学领域应用的不断深入,大量中医临床记录得以保留,为中医学研究学习提供了有效途径,而症状名识别是中医临床记录研究的重要前提。中医症状名识别可以看作一般文本中命名实体识别技术在中医学领域的特殊应用。我们采用命名实体识别中序列标注技术对中医临床记录进行研究,并结合中医临床记录领域特点,进行序列标注策略改进。通过HMM算法在序列标注策略改进前后实验结果对比,改进后HMM算法在性能评价指标上优于未改进之前,说明根据中医临床记录特点进行的序列标注策略改进是有效的。     

基于Pro/E的二维标注标准化和快捷化实现方法

针对Pro/E二维标注系统存在的不符合国标及效率低下等问题,提出了基于Pro/E二维标注系统的实现方案,并通过采用标准符号库、智能化尺寸标注、公差(配合)自动化查询以及简化人机交互等方法,实现了二维标注的标准化和便捷化.在基于Pro/E二次开发软件敏捷标注系统SmartDrawing上,对这些方法进行了验证.实例表明,采用SmartDrawing系统的二维标注,在标准化和快捷化等方面都有显著提升.     

基于棋盘空间的中文命名实体识别研究

命名实体识别NER是自然语言处理NLP领域中重要的基础任务之一。BIOE序列标注一般采用间接标注实体的方法,难以解决实体嵌套的问题;实体位置和实体类别两阶段的识别存在误差传播问题。针对该问题,根据实体在文中始末位置2个维度构建棋盘,再结合类别维度构建三维棋盘空间来标注实体,棋盘空间中的一个标注点即代表一个实体,是一种在高维空间直接标注实体的方法。该文在预训练模型BERT/NEZHA的基础上设计了一种NER模型,在CLUENER2020数据集上进行试验,棋盘空间标注的NER的F1指标分别为80.43和80.71,相比于BERT-CRF等模型略有提升,表明该标注方法和模型具备可行性和先进性。     

SOM-NCSCM+:抽取式神经网络中文标题生成方法研究

标题生成作为文本摘要任务的一个分支,能够帮助人们高效获取信息。本文针对中文标题生成任务面临的大规模、高质量中文标注数据缺乏的问题,利用标题往往可由原文中的词语来构成的特点,从将无监督学习模型与有监督的序列标注模型结合的角度出发,提出了融合聚类模型和主题模型的抽取式深度神经网络中文标题生成方法和模型。在缺乏人工分类标注信息的中文新闻数据集上,该模型可利用聚类和主题模型自动挖掘数据内部潜在的特征信息,获得不同的数据簇及各簇内的主题词来辅助中文新闻标题生成,使模型在具有潜在主题类别特征的、标题质量参差的中文新闻数据集上都具有较好的适用性。本文提出的中文标题生成模型在互联网上公开的中文新闻标题数据集上的实验结果也表明其在微观F1、BLEU、ROUGE、压缩率等评价指标上都取得了较基准模型更好的效果。     

利用类型语义表示进行标签降噪的细粒度实体分类

细粒度实体分类(FET)任务的训练数据往往利用已有知识库中的知识通过远程监督方法进行生成,生成过程中不可避免地引入多余的噪音标签。现有考虑训练数据中噪音问题的工作通常只建模训练数据和标注类型的概率分布,对细粒度类型的语义信息学习不足,造成在标注了多个细粒度类型的训练数据上选择了与实体上下文不相关的类型进行模型的学习。本文提出一种利用细粒度类型的语义表示进行标签降噪的细粒度实体分类方法。首先利用训练数据中具有唯一细粒度类型路径的数据学习一部分细粒度类型的表示,进而结合细粒度类型间的关系信息学习其他细粒度类型的表示;其次在标注了细粒度类型的训练数据中选取与实体上下文的语义信息最相似的细粒度类型为目标类型,从数据集中选择Top-K个相似数据进行细粒度类型语义信息的聚合;最后在聚合信息上学习最终的细粒度实体分类模型。实验结果表明,该方法可以有效地从标注了细粒度类型的训练数据中选出与实体上下文的语义信息最相符的细粒度类型,达到提升细粒度实体分类准确率的效果。     

基于TSVM与主动学习融合的蛋白质交互作用关系抽取

针对蛋白质交互作用关系(PPI)抽取研究中已标注语料有限而未标注生物医学自由文本易得的问题,进行了基于直推式支持向量机(TSVM)与主动学习融合的蛋白质交互作用关系抽取研究.通过自主选择最优的未标注样本加入到TSVM的训练过程中,最大程度地提高了系统的性能.实验结果表明,TSVM与主动学习融合的算法在少量已标注样本和大量未标注样本组成的混合样本集上取得了较好的学习效果,与传统的支持向量机(SVM)和TSVM算法相比,能有效地减少学习样本数,提高分类精度,在AImed语料上取得了F测度为64.12%的较好性能.     

Measurement of the thermal expansion of metal and FRPs

Thermal expansion is an important parameter for characterization of different binding forces, lattice dynamics, band and crystal structure of any solids. Many investigators have focused their attention to study this property theoretically and experimentally at different temperatures. It is one of the important properties of metals and its alloys, which helps to calculate the thermal stress. This parameter is also used to determine the compatibility of an insulator as load bearing materials. Different experimental setups have been developed to study thermal expansions of the materials using different techniques namely capacitance method, interferometric principle, LASER, optical, quartz tube etc. This paper reviews most of the experimental setups available to measure thermal expansion of metals, alloys, polymers and fibre-reinforced plastics at temperature ranging from 1 to 1100 K.     

Numerical investigation of folding of coated papers

Folding of coated paper is examined numerically using the finite element method. Particular emphasis is put on the behaviour of field variables relevant for cracking of the coating layers. In the numerical analysis, the basepaper is modelled as an anisotropic elastic–plastic material (both elastic and plastic anisotropy is accounted for) while the constitutive behaviour of the coating layers are approximated by classical (Mises) elastoplasticity. The numerical results suggest, among other things, that particular forms of plastic anisotropy can substantially reduce the maximum strain levels in the coating. It is also shown that delamination buckling, in the present circumstances, will have a very small influence on the strain levels in the coating layer subjected to high tensile loading.     

Analysis of optimum configuration of two-stage thermoelectric modules

In this paper, the theoretical analysis and stimulating calculation were conducted for a basic two-stage thermoelectric module, which contains one thermocouple in the second stage and several thermocouples in the first stage. The study focused on the configuration of the two-stage thermoelectric module, especially investigating the influences of some parameters, such as the allocation of the junction temperature difference in the module, the length of thermocouples and the number of thermocouples, on the cooling performance of the module. The obtained analysis results indicate that changing the junction temperature difference in the second stage, the length of thermocouples and the number of thermocouples in the first stage can improve the cooling performance of the module. These results can be used to optimize the configuration of the two-stage thermoelectric module, and provide guides for the design and application of thermoelectric cooler.     

As-cast microstructures and behavior at high temperature of chromium-rich cobalt-based alloys containing hafnium carbides

Hafnium is often used to improve the high temperature oxidation resistance of superalloys but not to form carbides for strengthen them against creep. In this work hafnium was added in cobalt-based alloys for verifying that HfC can be obtained in cobalt-based alloys and for characterizing their behavior at a very temperature. Three Co–25Cr–0.25 and 0.50C alloys containing 3.7 and 7.4 Hf to promote HfC carbides, and four Co–25Cr– 0 to 1C alloys for comparison (all contents in wt.%), were cast and exposed at 1200 °C for 50 h in synthetic air. The HfC carbides formed instead chromium carbides during solidification, in eutectic with matrix and as dispersed compact particles. During the stage at 1200 °C the HfC carbides did not significantly evolve, even near the oxidation front despite oxidation early become very fast and generalized. At the same time the chromium carbides present in the Co–Cr–C alloys totally disappeared in the same conditions. Such HfC-alloys potentially bring efficient and sustainable mechanical strengthening at high temperature, but their hot oxidation resistance must be significantly improved.     

Effect of a novel thixoforming process on the microstructure and fracture behavior of A356 aluminum alloy

In the present study, a novel thixoforming process for semi-solid deformation of A356 aluminum alloy is introduced using a continuous hot deformation process to the temperature being lower than the eutectic temperature of the alloy. A new hypothesis was introduced and the deformation mechanism of the alloy was investigated using the presented hypothesis. Microstructure and fracture surfaces of thixoformed samples were investigated using image analyzing technique and scanning electron microscopy. Obtained results indicated that this novel thixoforming process produces fine and compact silicon particles, dispersed uniformly in the microstructure of the alloy, compared to those produced by conventional thixoforming and gravity-cast processes with large and integrated morphology for silicon particles. The production stages of these silicon particles in this process were well documented by mentioned hypothesis. In order to investigate the effect of this novel process on mechanical properties of A356 alloy, tensile tests were conducted on produced samples. It was found that morphological changes of silicon particles as well as increasing the density ratio of samples in this process have a remarkable effect on enhancing the mechanical properties of produced alloy in comparison with other production routes. A new combination parameter, i.e. silicon density ratio (SDR) index was introduced. This parameter correlates the mechanical properties of samples to morphological properties of silicon particles and density ratio of them. Results of the study also indicated that samples with low SDR index have superior mechanical properties and consequently intergranular fracture mode.     

Percolation and the resistive transition of the critical temperature Tc of Nb3Sn

The relationship between the distribution of the critical temperature, the percolation function, and the resistive transition of the critical temperature is explored for polycrystalline Nb₃Sn. In the neighborhood of the critical temperature, Nb₃Sn is assumed to be a random mixture of superconducting and normal grains. Percolation concepts are applied to a study of the resistivity. A general analysis is made showing that the onset and shape of the resistive transition for composite conductors are determined by the percolation function and the distribution of the critical temperature. An approximate form of the percolation function is determined based on a linear FEM analysis. Example resistive transitions are calculated for an assumed normal distribution of the critical temperature. An argument is presented that relates grain orientation and strain dependence in Nb₃Sn. It is noted that a dependence of the distribution of T_c with strain, in addition to the usual shift in T_c with strain, would be the result of a strain dependence that is a function of grain orientation. The analysis shows the extent to which the slope of the resistive transition is a measure of the distribution of the critical temperature, and therefore a measure of the grain orientation strain sensitivity. Finally, a method is described to determine the percolation function experimentally.     

含能材料机械撞击感度判据的认识和发展

擅击感度"判据"的发展进程经历了上百年时间,由"爆炸概率"、"特性落高"到"爆炸的临界压力,炸药层的临界厚度"共经过了三个阶段.爆炸概率粗犷地区分各类炸药的擅击感度;特性落高可辨别每类炸药的撞击感度,尤其是高感度的炸药,如硝胺类化合物;而爆炸的临界压力和炸药的临界厚度则反映了在撞击作用下炸药发生快速反应的机理,可以更细微地分辨每种炸药的撞击感度.爆炸概率和特性落高适用于炸药撞击感度的常规检测,爆炸的临界压力和炸药的临界厚度适用于炸药撞击感度的理论研究.     

Effect of different treatments of copper surface on its total hemispherical absorptivity bellow 77 K

Total hemispherical absorptivity of copper surfaces treated with standard industrial methods was measured in dependence on the temperature of thermal radiation, varying from 25 K to 300 K. The sample temperature was typically from 5 K to 40 K and did not exceed 70 K. Usability of chemical and mechanical Cu surface finishing as well as Cu plating with Ni and Au for cryogenic design is discussed. As an example of practical application of our results, the cryogenic design of a LN₂ trap is presented.     

Effect of cryogenic treatment on microstructure, mechanical and wear behaviors of AISI H13 hot work tool steel

This paper focuses on the effects of low temperature (subzero) treatments on microstructure and mechanical properties of H13 hot work tool steel. Cryogenic treatment at −72 °C and deep cryogenic treatment at −196 °C were applied and it was found that by applying the subzero treatments, the retained austenite was transformed to martensite. As the temperature was decreased more retained austenite was transformed to martensite and it also led to smaller and more uniform martensite laths distributed in the microstructure. The deep cryogenic treatment also resulted in precipitation of more uniform and very fine carbide particles. The microstructural modification resulted in a significant improvement on the mechanical properties of the H13 tool steel.