面向电力多模态融合的语义差异性和感知能力差异性分析方法

多模态融合是实现电力系统数字化的重要技术手段,但多模态特征间的差异性限制了融合感知效果。因此,首先对电力多模态数据融合中的语义差异性和感知能力差异性现象进行了深入分析,对差异性产生的特征同化和权重决策问题进行了剖析。然后,针对语义差异性问题,使用角度差对多模态语义差异进行表征,并基于此寻找联合表征空间,实现电力多模态特征同化;其次,针对感知能力差异性问题,使用交叉损失熵对电力多模态感知能力进行表征,并基于此构建权重决策模块,实现多模态特征融合权重计算。最后,以前期所提融合框架为基础,提出了针对多模态差异性问题的高容错性特征融合模型。仿真以输变电线路应急抢修场景为例,基于所提分阶段训练策略进行模型训练,并从融合感知、特征同化以及权重决策机制3个方面验证了所提方法的有效性。

动补结构“语义差”标准例说

我们考察动补结构上下字的关系和汉语史语料发现,动补结构上下字之间存在"语义差":上字的行为动作性质强于下字,下字的状态结果性质强于上字,上字的行为动作产生了下字的状态结果,该文通过剖析"澄清"等例证来说明动补结构的形成机制。

基于词语语义差异性的多标签罪名预测

罪名预测是智慧司法领域中的一项重要研究内容,其旨在依据犯罪事实自动预测出犯罪主体触犯的罪名。犯罪事实是案件的真实客观描述,犯罪事实中各词语的语义重要性在不同罪名的判决中有所差异,而现有方法在对犯罪事实建模的过程中往往忽略了这种语义差异性,且缺乏对数罪并罚情形的处理。为此,该文在对犯罪事实的建模过程中将词语的语义差异融入注意力机制;并将数罪并罚情形下的多标签罪名预测转化为多个独立的单标签罪名预测。实验结果表明,该文基于词语语义差异性建模和多标签转化策略均有利于提升罪名预测的效果,在“中国法研杯”2018司法人工智能挑战赛公布的数据集上达到了88.0%的F1值。

基于双通道语义差网络的方面级别情感分类

方面级别情感分类旨在分析一个句子中不同方面词的情感极性。先前的研究在文本表示上,难以产生依赖于特定方面词的上下文表示;在语义特征分析上,忽略了方面词的双侧文本在整体语义上与方面词情感极性之间具备不同关联度这一特征。针对上述问题,该文设计了一种双通道交互架构,同时提出了语义差这一概念,并据此构建了双通道语义差网络。双通道语义差网络利用双通道架构捕捉相同文本中不同方面词的上下文特征信息,并通过语义提取网络对双通道中的文本进行语义特征提取,最后利用语义差注意力增强模型对重点信息的关注。该文在SemEval2014的Laptop和Restaurant数据集以及ACL的Twitter数据集上进行了实验,分类准确率分别达到了81.35%、86.34%和78.18%,整体性能超过了所对比的基线模型。

Whether和If语义句法差研究

既往的whether、if比较研究,无论是从whether、if的wh语义特征出发,还是从whether、if句法结构差异出发,都没有很好解决whether、if的句法语义差,特别是whether、if在句法分布上的差异。鉴于此,在区别whether表语义可选择及if表语义无选择的差异基础上,两者在引导条件句和正反疑问句时的结构和分布差异被重新梳理。特别是在引导正反疑问句时,whether的DP属性由于句法匹配、句法属性管辖等原因,使其能够在较多的句式结构中出现;而if的连词属性及引导结构的附加语成分属性极大限制了它的句法分布。

特殊话语标记和语义无差异性——论加缪《局外人》与塞林格《麦田里的守望者》的叙事意义

本文从加缪的《局外人》和塞林格的《麦田里的守望者》的特殊话语标记入手 ,研讨作品的叙事特征和意义 ,指出它们作为小说叙事文本的最大特点就是揭示了以互不相容的语义价值和文化价值拼接起来的双重性 ,导致了价值的无差异性。作品中大量表示折衷立场的词语和语义对应了特定社会的深层结构组织状况和战后社会语言环境因意识形态冲突而导致的堕落。

融入法因层次结构的法因预测IHLCP模型

该文针对当前法律智能体系可解释性差、低频易混淆法因预测效果不佳、民事纠纷研究过少的问题,设计了一种可解释性层次法因预测IHLCP模型,并将法因之间的层次依赖关系作为模型可解释性的来源进行了研究。模型首先基于案件的语义差异性对事实描述进行编码,然后通过改进的seq2seq-attention模块来预测法因路径,并利用法因内部的文本信息过滤事实描述中的噪声信息,以获得可靠的预测效果。该文设计的IHLCP模型在CIVIL、FSC和CAIL这三个大规模公开数据集上分别达到了当前最好的效果(CIVIL数据集:ACC-91.0%,PRE-67.5%,RECALL-57.9%,F1-62.3%。FSC数据集:ACC-94.9%,PRE-78.8%,RECALL-75.9%,F1-77.3%。CAIL数据集:ACC-92.3%,PRE-90.9%,RECALL-89.7%,F1-90.3%),其中ACC和F1值分别最高提升了6.6%和13.4%。实验结果表明,该设计能够帮助系统理解法因,弥补了当前法律智能体系在低频、易混淆法因预测上的不足,同时提升了模型的可解释性。

上古汉语动补结构判别标准的讨论

王力先生提出了从形式和意义两方面来判断动补结构的标准,但是,形式和意义两个标准必须有机地统一。在意义方面,应该从上下字的关系上考察,而不是着眼于上字或下字的词汇意义是否虚化,语法上是否及物。通过考察动补结构上下字的关系可以发现,动补结构上下字之间存在"语义差":上字的行为动作性质强于下字,下字的状态结果性质强于上字,上字的行为动作产生了下字的状态结果。这种上下字之间的"语义差"可以作为判别古汉语动补结构的一个主要标准。

基于词向量与TextRank的关键词提取方法

针对词汇语义的差异性对TextRank算法的影响进行了研究,提出一种基于词向量与TextRank的关键词抽取方法。利用FastText将文档集进行词向量表征,基于隐含主题分布思想和利用词汇间语义性的差异,构建TextRank的转移概率矩阵,最后进行词图的迭代计算和关键词抽取。实验结果表明,该方法的抽取效果相比于传统方法有明显提升,同时证明利用词向量能简单而有效地改善TextRank算法的性能。

文化资源型旅游城市形象提升设计研究:以江苏千灯古镇为例

目的 以江苏省苏州市昆山市千灯古镇为例,提出一种文化城市形象研究与再设计的方法。方法 通过IPA模型分析城市形象和城市具体名片的组成部分的重要程度和满意度;通过主成分分析模型分析成分具体的影响因子;通过FAHP量化需求指标,得出需求权重。结果 千灯古镇旅游特色产品开发中,昆曲艺术风格综合权重为0.336,纪念性物品综合权重为0.171,特色古镇建筑风格综合权重为0.146;调研的20~32岁年龄段的中青年群体更加偏向文化特征性较强的旅游纪念品。结论 城市形象的提炼可以通过模型方法挖掘关键元素和设计表达符号,该设计方法有利于城市形象建设和提升。

Reducing semantic bias of annotations for semantic web service discovery

In order to improve the effectiveness of semantic web service discovery, the semantic bias between an interface parameter and an annotation is reduced by extracting semantic restrictions for the annotation from the description context and generating refined semantic annotations, and then the semantics of the web service is refined. These restrictions are dynamically extracted from the parsing tree of the description text, with the guide of the restriction template extracted from the ontology definition. New semantic annotations are then generated by combining the original concept with the restrictions and represented via refined concept expressions. In addition, a novel semantic similarity measure for refined concept expressions is proposed for semantic web service discovery. Experimental results show that the matchmaker based on this method can improve the average precision of discovery and exhibit low computational complexity. Reducing the semantic bias by utilizing restriction information of annotations can refine the semantics of the web service and improve the discovery effectiveness.

现代语文词典应拒绝同义相训

从语言的运用出发 ,提供各种有效信息帮助读者掌握词语的意义和用法 ,并使之最终用好词语 ,这已成为现今世界语文词典的既定目标。同义相训虽然通用已久 ,但有碍这一目标的实现。因为它只能利用解释词与被解释词在意义上的相似对词语作出粗略解释 ,无法说明后者的意义内容及其个性特点 ,更难以揭示其词义与用法的内在联系 ;此外 ,同义相训忽略同中之异 ,易使读者认为解释词与被解释词之间毫无差别 ,从而造成误解和误用。我国 80年代以来出版的词典曾试图缩小同义相训的副作用 ,但补救的办法毕竟不能消除同义相训自身的局限。从编纂宗旨和释义的要求来看 。

基于大数据环境下面孔空想性错视对产品造型影响研究

目的探究面孔空想性错视现象是否会对产品造型的审美满意度产生影响。方法采用实证研究的方式,运用大数据网络爬虫程序提取产品的语义信息、通过词频分析方法构建语义差分量表,并分别建立三组对照组实验对被试者进行问卷调查,最后分析三组实验数据得出研究结论。结果实验发现被试者在不考虑产品价格、功能的情况下,审美偏好更倾向于能够引起面孔空想性错视的产品造型。结论在产品造型设计中运用面孔空想性错视现象,可以在一定程度上引起网购消费者的注意力,在传达产品情绪化特征的同时,满足其功能和审美需求,从而对产品造型审美满意度产生正向影响。运用面孔空想性错视现象是满足用户需求、提高网购消费者购买意愿的有效设计途径和方法。

基于SD法的老年人室外公共活动空间使用评价研究——以武汉理工大学西院校区为例

以位于武汉市洪山区的武汉理工大学西院老校区内三处老年人室外公共活动场所为研究对象,通过对这三处老年人活动场地进行前期调查,探究老年人室外公共休闲活动空间的影响因素及其影响程度。在环境行为与心理学原理的基础上,结合语义差分析法及统计分析法,定量研究老年人对室外公共休闲活动空间的使用评价,总结现有老校园内老年人公共活动场所的特点,揭示老年人室外休闲环境的影响因素,其中公共活动场所的环境因子(F1)和位置因子(F2)对老年人的使用意愿影响最大。最后,基于上述客观的调查与分析,对本校区内老年人的室外活动场所提出针对性的空间营造建议。

An Explorative Research into the Factors Affecting Chinese Students＇ Retention and Application of English New Words

Vocabulary has a direct impact on people＇s communication competence. A good mastery of an English word covers its pronunciation, spelling, part of speech, meaning and application. This article will focus only on factors affecting the retention and application of a word from the perspective of semantic disparities, different defining angles in English and Chinese, emotional coloring and styles, morphological changes, and collocation. The author hopes that this article will be able to provide some insights into vocabulary teaching.

Semantic Differential as an Assessment Tool of （Dis） Advantages of QMS in the Light of Accredited Certification in Poland

This article is a continuation of the issue in the field of assessment of （dis）advantages resulting from the implementation of the quality management systems （QMS） in a structure of the organization. The publication uses the results of studies conducted in the range of assessment＇s information completeness provided by the certification bodies （CBs） in the context of advantages and disadvantages of the QMS implementation in accordance with PN-EN ISO 9001. In addition, an attempt to define a tool for the QMS assessment of （dis）advantages in the organization has been made. The basis for empirical analysis of the issue was the information provided by an accredited management systems CBs recommended by the Polish Centre for Accreditation （PCA）. There are 33 units in Poland, which were granted by PCA with the accreditation within the range of ISO 9001：2008. In the first stage of the research work, a substantive analysis of websites content among mentioned CBs has been carried out. In addition, the paper proposes use of semantic differential method （SD） to evaluate （dis）advantages of the QMS. SD adaptation for the assessment of the QMS （dis）advantages is the authorial concept. The authors hope for polemic in this area. In the face of obtained results, an authorial SD construction has been proposed, which is useful and may help in the formulation and conveying of the relevant （reliable） information to potential clients in the range of （dis）advantages resulting from the QMS implementation to the organization. As a consequence, the recommendations have been set, which are kind of guidelines that could significantly influence the realignment of the information state conveyed on the internet by the CBs.

Affective Meaning of Pronouns in a Russian Sample

The present study examines affective meaning of pronouns （in Russian） represented by the semantic differential. Of direct relevance to the present study is the theory of affective meaning propounded by Osgood. Two hypotheses were tested. According to a ＂magnitude＂ hypothesis, affective dimensions （each of evaluation, potency, and activity taken separately） differ in their magnitude across pronouns I, My, Me, and They. A ＂structural＂ hypothesis stated that the affective dimensions yield latent factors across （the generality） and within （the concept-scale interaction） the pronoun concepts. Repeated-measures analysis of variance （1-way ANOVA） and confirmatory factor analysis were employed to process data. It was found that each of the evaluation and potency measures yield a significant magnitude change across the pronouns, but there was indicated no significant change across the pronouns with respect to the activity dimension. Therewith, the pronoun My gained a salient value and the pronoun They the smallest value. Using confirmatory factor analysis five models were tested. Among them one model was good fit to the data. It engaged a four-factor solution resulted in that four pronouns are latent affective distinct but related factors and the evaluation, potency, and activity are their indicators.

Definition of Standard English and its enlightenment on English teaching

Standard English is defined not as a language, a style, an accent, but as standardized grammar and vocabulary with different accents. And from a linguistic point of view, Standard English is no more correct than any other form of English. Yet it is most prestigious, and a desirable educational target. However, different varieties of English should also be taught for international communication.

顾及空间及语义相关性的POI位置标注优化方法

POI的质量是确保寻路、路径规划、导航等高质量地图服务的基础,是影响POI数据的位置准确性、逻辑一致性和语义差异性问题等数据质量问题的关键因素。本文从影响POI位置空间、关系和语义因素3个角度出发,针对POI位置错标、不同语义POI无法区分及POI分布与现实世界不一致的情况,结合建筑数据和道路数据,进行POI多阶段位置优化研究,包括位置约束计算阶段、位置选择阶段、位置优化阶段。以某街道区域为分析区域开展POI位置优化实验,结果表明本文提出的位置优化方法能有效降低错标率(19.3%)、提高可辨识度(6.7%),并有效提高子区域POI分布与道路分布的一致性,有效提高不同语义POI的识别速度。

Discriminant Models for Uncertainty Characterization in Area Class Change Categorization

Discriminant space defining area classes is an important conceptual construct for uncertainty characterization in area-class maps.Discriminant models were promoted as they can enhance consistency in area-class mapping and replicability in error modeling.As area classes are rarely completely separable in empirically realized discriminant space,where class inseparabil-ity becomes more complicated for change categorization,we seek to quantify uncertainty in area classes(and change classes)due to measurement errors and semantic discrepancy separately and hence assess their relative margins objectively.Experiments using real datasets were carried out,and a Bayesian method was used to obtain change maps.We found that there are large differences be-tween uncertainty statistics referring to data classes and information classes.Therefore,uncertainty characterization in change categorization should be based on discriminant modeling of measurement errors and semantic mismatch analysis,enabling quanti-fication of uncertainty due to partially random measurement errors,and systematic categorical discrepancies,respectively.