基于图卷积神经网络的中文实体关系联合抽取

现有实体关系联合抽取方法未充分考虑中文句子中实体关系的复杂结构特征,为此,提出一种基于图卷积神经网络（GCN）的中文实体关系联合抽取方法。在双向长短时记忆网络抽取序列特征的基础上,利用GCN编码依存分析结果中的语法结构信息,借鉴改进的实体标注策略构建端到端的中文实体关系联合抽取模型。实验结果表明,该方法的F值可达61.4%,相比LSTM-LSTM模型提高了4.1%,GCN能有效编码文本的先验词间关系并提升实体关系抽取性能。     

Fodorian semantics,pathologies, and “Block's Problem”

In two recent books, Jerry Fodor has developed a set of sufficient conditions for an object X to non-naturally and non-derivatively mean X. In an earlier paper we presented three reasons for thinking Fodor's theory to be inadequate. One of these problems we have dubbed the Pathologies Problem. In response to queries concerning the relationship between the Pathologies Problem and what Fodor calls Block's Problem, we argue that, while Block's Problem does not threatenFodor's view, the Pathologies Problem does.We would like to thank Ray Elugardo, Pat Manfredi, and Donna Summerfield for helpful comments on an earlier paper on Fodorian Semantics, X means X: Semantics Fodor-Style. We would especially like to thank Ned Block for extended e-mail conversations about Block's Problem. Block agrees that his problem is not the same as our pathologies problem. Contrary to what we say here, he still maintains that his objection can ultimately be made to work to defeat Fodor's theory of meaning. His elaboration of Block's Problem is different than the one we present here. Versions of a related paper were presented at the 1991 Annual Meeting of the Southern Society for Philosophy and Psychology as well as the Canadian Society for History and Philosophy of Science.     

向上学习方法改进移进归约中文句法分析

基于移进归约的句法分析系统具有线性的时间复杂度,因此在大规模句法分析任务中具有特别实际的意义。然而目前移进归约句法分析系统的性能远低于领域内最好的句法分析器,例如,伯克利句法分析器。该文研究如何利用向上学习和无标注数据改进移进归约句法分析系统,使之尽可能接近伯克利句法分析器的性能。我们首先应用伯克利句法分析器对大规模的无标注数据进行自动分析,然后利用得到的自动标注数据作为额外的训练数据改进词性标注系统和移进归约句法分析器。实验结果表明,向上学习方法和无标注数据使移进归约句法分析的性能提高了2.3%,达到82.4%。这个性能与伯克利句法分析器的性能可比。与此同时,该文最终得到的句法分析系统拥有明显的速度优势(7倍速度于伯克利句法分析器)。     

Supporting concept location through identifier parsing and ontology extraction

Identifier names play a key role in program understanding and in particular in concept location. Programmers can easily “parse” identifiers and understand the intended meaning. This, however, is not trivial for tools that try to exploit the information in the identifiers to support program understanding. To address this problem, we resort to natural language analyzers, which parse tokenized identifier names and provide the syntactic relationships (dependencies) among the terms composing the identifiers. Such relationships are then mapped to semantic relationships.     

Learning linear non-Gaussian networks: A new view from matrix identification

Identifying causal structures from observations is fundamental in many applications. Probabilistic graphical models provide a unifying framework for capturing complex causal dependencies among random variables. Recently, a linear non-Gaussian acyclic model (LiNGAM) and some smart algorithms (ICA-LiNGAM, DirectLiNGAM) have been proposed, which outperform previous graphical models and learning methods in identifying variable orders. We propose new solutions (TMFLiNGAM, SchurLiNGAM and RCLiNGAM) to the LiNGAM learning task from the perspective of matrix identification. TMFLiNGAM and SchurLiNGAM recover orders more directly, and RCLiNGAM can improve the accuracy of previous algorithms on uniform and sparse structures. The perspective also facilitates the learning of sparse models where the performance of all independent component analysis-based algorithms can be improved by reconstructing variable orders from the inverse of separation matrices. Experimental results under various settings provide average evaluations over the learning methods, and verify the effectiveness of our perspective and algorithms.     

基于粗集优势关系的属性赋权相对熵优化模型

针对传统粗集理论中属性赋权不一致,甚至相悖的问题,把2个概率分布的相对熵扩展到任意2个单维向量的相对熵,并将相对熵视作一种距离。通过定义属性重要度的代数观和粒度观确定优化权重的取值范围,根据各方案的属性值尽可能靠近理想值、远离负理想值的原理,建立单目标赋权优化模型。针对等价关系的局限性,将优势关系引入属性权重确定方法中。基于优势关系的序信息系统,将代数观下和粒度观下的权重通过相对熵优化模型进行耦合,得到多属性决策中属性权重的优化解。算例分析结果证明了该模型的有效性。     

自然语言向SQL代码的转化方法

为解决智能学习系统查询语言的转化问题,提出一种自然语言向SQL代码转化的方法。利用所建立的字典扫描单词和理解语义,采用改进后的单词提取技术扫描自然语言串,以生成语义依赖树,并将其语义关系划分为若干独立的集合块,通过对该集合块遍历生成与自然语言等价的SQL代码。实验结果表明,该转化方法简单有效。     

基于树核函数的中文语义角色标注

使用SVM提供的卷积树核函数构造一个中文语义角色标注系统,将依存关系作为标注单元进行中文语义角色标注。通过不同的裁剪方法获得依存树的结构化信息,裁剪后的依存树分别为最短路径树和最小树。在中文PropBank和NomBank的转换语料上进行实验,结果表明,该系统在动词性谓词和名词性谓词语料上的F1值分别为83.66和76.87。     

基于子树的确定性依存分析方法

确定性依存句法分析(又称为基于动作的依存句法分析或基于转换的依存句法分析),由于采用了贪婪的搜索策略,它是一类比较有效率的依存句法分析方法.在一种主流的确定性依存句法分析方法-Yamada-Matsumoto算法的基础上,通过改进其转换模型,提出了一种基于子树的确定性依存句法分析方法.在新的方法中,引入了子树的结构,因而能够更加灵活地确定词与词间的依存关系.最后,通过在中文宾州树库上的实验,证明了这个新方法的有效性.     

基于相关性模型的电路板TPS开发与验证

用TPS对电路板进行故障检测和维修,是提高新装备维修保障能力的重要环节.对TPS的开发者和最终使用者而言,如何提高TPS开发效率,并对其进行验证已经成为亟需解决的问题.分析了几种TPS验证方法的特点,提出了基于相关性模型的电路板TPS开发和验证方法.从测试性模型建立、关联矩阵生成和诊断策略设计等方面阐述了该过程.对于面...