一种基于规则的神经网络的知识表示和推理方法

在医学专家系统中，如何根据各种疾病特征，推断出患某种疾病，这是一个比较复杂的问题，本文给出了一种基于规则的人工神经网络的知识表示和知识推理方法，从规则集中自动构造网络的结构、确定隐层节点数和连接权值，用并行的方法进行推理，最后，给出了泌尿外科常见疾病诊断的一个实例应用。     

基于本体的领域知识表示和获取

本体的方法是新的知识表示的有效手段。文章讨论分析了利用RDR和FCA对领域知识表示和获取,以及在此基础上的本体的构建方法。探讨了在这种机制下如何实现对领域信息的抽取。     

基于概念图的关联规则知识表示

针对传统关联规则表示方式无法展现领域知识、数据项间的关系及规则中所隐含的信息等问题,提出了一种基于概念图的关联规则知识表示方法,该方法包括模式定义和模式解析,其结合概念图理论可将关联规则转换成概念图的知识表示形式。给出了关联规则的概念图知识表示算法,并以某省全员人口数据为数据源对算法进行了具体实现和分析。实验结果表明,该方法在人口信息表现方面具有良好的效果。     

锅炉故障诊断专家系统领域知识表示方法研究

选择什么样的知识表示方法以及如何恰当地把领域知识表示成工程知识,是专家系统开发中必须面对的问题。该文以电站锅炉的领域知识为研究对象,首先讨论了知识的分类及其表示方法,再根据各种表示方法的特长与不足,并结合锅炉领域知识的特征,选用产生式表示法和框架表示法进行知识表示,最后以锅炉的水冷壁管爆破为例,介绍如何把领域知识表示为工程知识。     

知识表示学习研究进展

人们构建的知识库通常被表示为网络形式,节点代表实体,连边代表实体间的关系.在网络表示形式下,人们需要设计专门的图算法存储和利用知识库,存在费时费力的缺点,并受到数据稀疏问题的困扰.最近,以深度学习为代表的表示学习技术受到广泛关注.表示学习旨在将研究对象的语义信息表示为稠密低维实值向量,知识表示学习则面向知识库中的实体和关系进行表示学习.该技术可以在低维空间中高效计算实体和关系的语义联系,有效解决数据稀疏问题,使知识获取、融合和推理的性能得到显著提升.介绍知识表示学习的最新进展,总结该技术面临的主要挑战和可能解决方案,并展望该技术的未来发展方向与前景.     

基于知识树的领域知识组织和应用*

在分析已有知识表示方法优缺点的基础上,提出一种高效的知识表达模型——概念知识树。概念知识树模型不仅结构性好、表达能力强,而且在应用中具有良好的适应性和延展性,现主要应用于信息检索和自然语言理解领域。以旅游领域为背景,用概念知识树作为表达模型建立相应的领域知识体系,并在此基础之上实现了一个旅游智能分析系统。该系统可在对用户需求(自然语言)进行一定程度语义理解的基础上,根据所建立的知识体系自动进行语义匹配,最终返回满足用户需要以及相关的旅游路线信息。     

关系数据库表示规则知识的理论与方法

知识库及其组织结构在智能系统中具有相当重要的作用。针对当前知识库的难点问题,结合实际应用,给出了一种基于关系型数据库的简便通用的知识库结构设计,利用表记录项中的复合数据表示复杂信息,同时,提出了一种编码方法来保证知识的管理与检索。最后给出了一个实例,表明该方法可以完善地表达事实——规则知识体系,具有很好的普适性。     

知识的单形表示方法研究

本文基于形式逻辑的“三段论”以及代数拓扑学的三角剖分理论提出了一种新的知识表示方法——单形表示法,并与其它一些知识表示方法进行了比较．它在表达能力、表达方式及解释能力等方面都有明显的特点．     

三值关联规则在不确定性知识表示及推理性的应用

动态分类是指对象所属的类可随时间而变化。在进行系统分析和软件开发中,需要真实地描述研究对象在其生命周期中的变化,动态分类正是这样一种机制。但是由于许多面向对象程序设计语言都不支持动态分类,因此需要用别的方法来描述动态分类,文中给出了用ＵＭＬ表达动态分类的四种机制,并对这四种机制进行了比较。     

基于外部知识和层级篇章表示的阅读理解方法

阅读理解指的是基于给定文章自动回答相关问题,这是人工智能及自然语言处理领域的一个研究热点。目前已提出许多基于深度学习的阅读理解方法,但是这些方法对问题理解及篇章建模不充分,导致模型获取答案准确率不高。为了解决上述问题,该文提出一个基于外部知识和层级篇章表示的阅读理解方法。该方法特点有: ①通过引入问题重要词的字典释义、HowNet义原,并结合问题类型,加强问题理解; ②使用层级篇章表示,提升模型对篇章的理解; ③在一个框架下联合优化问题类型预测与答案预测两个子任务。在DuReader数据集上的实验结果表明,该方法与基线系统性能相比最大提升了8.2%。     

一种用于人脸识别的contourlet域稀疏表示分类器

超完备稀疏表示法可以有效解决人脸识别领域中由于光照、表情变化、遮挡和噪声问题等原因造成的性能瓶颈.基于超完备稀疏表示法,将人脸识别问题看作是为多个线性回归模型中的分类问题,提出了一种基于Contourlet域的稀疏表示分类器,改进了利用主成分分析进行数据预处理所造成的鉴别信息丢失,提升了稀疏表示分类器的鉴别能力.在ORL库、Yale库、扩展Yale库和PIE库上大量实验结果验证了算法的有效性.     

一种用于航天器数据监视的知识表示方法

针对我国目前以人为主的航天器数据监视现状,利用人工智能的一般方法,分析了航天器数据监视中需要用到的各种监视知识,提出一种将各种复杂监视知识表示成计算机可识别形式的方法;该方法采用基于规则的思想,易于接受,并引入了优先级的机制完成冲突消解;该方法能够解决与时间有关,与历史数据有关、基于曲线变化等几类遥测参数的监视问题,具有一定的实用性和推广价值。     

Analog and Digital Representation

In this paper, I argue for three claims. The first is that the difference between analog and digital representation lies in the format and not the medium of representation. The second is that whether a given system is analog or digital will sometimes depend on facts about the user of that system. The third is that the first two claims are implicit in Haugeland's (1998) account of the distinction.

On the Logical Foundations of Compound Predicate Formulae for Legal Knowledge Representation

In order to represent legal knowledge adequately, it is vital to create a formal device that can freely construct an individual concept directly from a predicate expression. For this purpose, a Compound Predicate Formula (CPF) is formulated for use in legal expert systems. In this paper, we willattempt to explain the nature of CPFs by rigorous logical foundation, i.e., establishing their syntax and semantics precisely through the use of appropriate examples. We note the advantages of our system over other such systems and discuss the significance of CPFs with regard to the formalization of legal reasonings using examples from the United Nations Convention for the International Sale of Goods.     

Argument Based Machine Learning Applied to Law

In this paper we discuss the application of a new machine learning approach – Argument Based Machine Learning – to the legal domain. An experiment using a dataset which has also been used in previous experiments with other learning techniques is described, and comparison with previous experiments made. We also tested this method for its robustness to noise in learning data. Argumentation based machine learning is particularly suited to the legal domain as it makes use of the justifications of decisions which are available. Importantly, where a large number of decided cases are available, it provides a way of identifying which need to be considered. Using this technique, only decisions which will have an influence on the rules being learned are examined.     

Brave Mobots Use Representation: Emergence of Representation in Fight-or-Flight Learning

The paper uses ideas from Machine Learning, Artificial Intelligence and Genetic Algorithms to provide a model of the development of a fight-or-flight response in a simulated agent. The modelled development process involves (simulated) processes of evolution, learning and representation development. The main value of the model is that it provides an illustration of how simple learning processes may lead to the formation of structures which can be given a representational interpretation. It also shows how these may form the infrastructure for closely-coupled agent/environment interaction.     

Awareness, Representation and Interpretation

This paper discusses how representation andinterpretation affect the degree and character of awarenessafforded by computer systems: awareness of people and ofinformation artifacts. Our discussion ranges from system designto theoretical concepts, and we focus on consistencies acrossthis spectrum. We begin by briefly describing a prototypecollaborative filtering system, Recer. This system tracks ongoingactivity in the web browsers and text editors of a group ofpeople, and offers recommendations of URLs and local programfiles that are specific to and adaptive with that activity, andthat reflect patterns of earlier activity within the community ofuse. We then take a more general look at collaborative filtering,and compare it with two other approaches to engendering awarenessof useful artifacts: information retrieval and software patterns.We discuss how each implicitly or explicitly involvescollaboration, formalisation and subjectivity in its corerepresentations. We then explore the artifact-centred approach toawareness that Recer represents, and relate it to theactivity-centred approach more familiar within CSCW. We use thiscomparison in discussing, in more theoretical terms, howrepresentation and formalisation affects awareness,interpretation and use. Our intention is to explore andunderstand the choices that designers have for the corerepresentations of information systems, and the consequences forawareness that follow for users. We wish to relate such practicaldesign issues to the more theoretical discussion in CSCW aroundconcepts such as common information spaces, the space-placedistinction, and the status of formal constructs.     

Representation Operators and Computation

This paper analyses the impact of representation and search operators on Computational Complexity. A model of computation is introduced based on a directed graph, and representation and search are defined to be the vertices and edges of this graph respectively. Changing either the representation or the search algorithm leads to different possible complexity classes. The final section explores the role of representation in reducing time complexity in Artificial Intelligence.