基于上下文的知识表示和推理--人工智能的观点

本文中,我们从知识表示和推理(KRR)的角度概括地阐述了上下文推理的概念和基本原理。首先阐述了上下文的概念;然后介绍了上下文推理的三种基本形式和上下文理论的两个基本原理,也就是局部性原理和一致性原理;接着讨论了上下文推理的形式化问题;最后通过对一个叫“魔术盒问题”的求解来展示如何利用多上下文系统MCS对问题进行形式化表示和求解。     

信息集成中上下文知识的描述与推理

基于本体的语义信息集成主要解决了异构数据源之间的模式级异构,但对于分布环境下普遍存在的上下文异构无法解决,这导致用户从信息集成系统得到的结果仍然存在语义异构。针对这种不足,本文提出了一种上下文知识的形式化描述和推理机制,并将这种上下文机制引入基于本体的语义信息集成中,同时对其进行扩展,使得扩展后的语义信息集成系统可以自动检测和消除上下文语义异构,从而形成了完整的语义异构解决方案。     

基于粗糙FCA-概念代数的上下文本体建模

在上下文本体模型中,根据现有上下文信息推导出新知识,但在推理过程中存在两个问题:(1)现有上下文中可能隐含多个有用信息,而现有方法在推理前并未对其针对这一点进行处理,上下文具有不完整性,推理出的知识可能不全面;(2)推理后有新知识出现,新知识与旧知识可能存在不协同等问题,使得本体可能没有较好的可扩展性。针对以上两个问题,借鉴粗糙FCA的粗糙处理方法,提出基于粗糙FCA上下文抽取方法以获得隐含上下文;再使用概念代数将得到的所有上下文深度形式化表示,并构建具有较好可扩展性的概念网。实验结果表明,在提出的方法基础上进行上下文推理的正确率高于直接使用原推理方法,而且在本体可扩展性方面有明显优势。     

基于交互上下文的预测方法

传统的上下文预测是在单用户的上下文基础上进行的,忽视了实际普适计算环境中由于用户交互活动导致的上下文变化因素。为了合理、有效地解决上述局限性问题,该文提出基于交互上下文的预测,给出交互上下文及交互上下文预测的概念,以及相关的建模、推理算法。     

语义信息集成中格式异构解决方案

基于本体的语义信息集成能够解决分布环境下异构数据源之间的模式异构,而对于广泛存在的上下文异构却无法解决。由于上下文异构是暗含的语义,无法为信息系统俘获和理解,要解决上下文异构,必须将上下文语义进行形式化描述。本文首先提出了一种将暗含的上下文语义进行形式化描述的方法,然后在此基础上提出了一种基于元数据格式表示的上下文转换方法来解决上下文异构中的格式异构。该方法避免了已有转换方法需要反复定义大量映射的缺点,提高了上下文转换的灵活性、适应性和扩展性。     

上下文感知在Smart Home系统中的应用

针对Smart Home环境的复杂性问题,提出了利用上下文技术来提高其智能化的方法.阐述了在普适环境下的上下文在智能家居环境中的重要性,介绍了上下文和上下文感知计算的定义,采用了基元组的形式定义了不同的上下文信息;提出了基于上下文感知的Smart Home系统模型,并对其功能进行了详细的描述;在Smart Home系统中,应用该模型对具体的上下文实例进行了描述和推理.     

基于觉察上下文计算的健康智能家庭研究

利用觉察上下文计算技术来研究实现健康智能家庭,主要研究了健康智能家庭的上下文建模和上下文推理,并构建了一个实验系统AngelHome,分析了健康智能家庭中的各种上下文信息,利用本体技术对其进行建模,并用OWLDL语言表达上下文信息模型,构建了AngelHome本体;在上下文推理部分采用混合推理,对不同的推理任务分别采用本体推理机、自定义规则推理机和贝叶斯神经网络推理.AngelHome采用OSGi框架,具有良好的伸缩性,这里分析了系统的几个主要部分,并进行了测试.实验结果表明,利用觉察上下文计算技术来实现健康智能家庭是可行的.     

基于描述逻辑的上下文知识获取与推理方法

针对上下文感知计算中缺乏清晰统一的模型与自动推理支持的问题,提出一种基于描述逻辑的上下文知识获取与推理方法。该方法首先提出了一种本体引导的上下文模型框架,根据抽象层次的不同将上下文模型分为元模型与领域特定模型两层结构;然后采用描述逻辑表示语言SHOIN(D)形式化描述该上下文模型,设计上下文模型向描述逻辑知识库的转换算法。最后以一个实际案例说明该方法的可行性。     

支持上下文感知应用程序的动态自适应中间件框架

上下文感知计算是当前开发和部署智能应用不可或缺的关键技术之一。上下文能否在计算中真正发挥其作用,主要取决于两方面：一是如何连续稳定地从动态交互环境中获取高质量上下文,二是如何推理上下文并制定适应决策。为了实现上述目标,设计了一个分层的中间件框架,该中间件能够根据上下文质量参数,动态地选择能提供高质量上下文的信息源,并对这些原始上下文进行预处理和推理,进而自动地制定适应决策为用户提供合适的服务。实验测试了平台的性能,并与同类系统进行了比较,结果表明该中间件能够快速有效地支持上下文感知应用的开发部署,并且在计算性能方面有显著提高。     

普适计算环境下上下文信息推理的研究

从各种低层上下文信息得到对人们更加有用的高层上下文信息即上下文推理是当前研究的热点.针对该问题,采用描述逻辑,研究基于本体模型的上下文推理方法.首先简要介绍基于本体的上下文模型,该模型增加了对上下文特性的建模,然后分别研究基于本体的推理、基于规则的推理及不一致性验证3种推理方式,借助Jena框架的推理接口实现,推理功能全面,通用性强,基本满足了普适计算系统中上下文推理的需求,最后给出了推理的可用性.     

Towards a Knowledge-Level Model of Context and Context Use in Diagnostic Problems

Diagnostic support systems that help solving problems in open and weak theory domains need to be context-sensitive in order to reveal flexible and efficient behaviour. This paper presents a task-oriented methodology for analysing and modeling contextual knowledge at the knowledge level. We present a context-sensitive diagnosis approach (ConSID) which clarifies the connection between content and process knowledge. The former embodies the domain model, while the latter embodies the task and method models. We present a prototypical system, the ConSID-Creek, that applies the ConSID approach to the medical diagnostic domain. We illustrate how the system integrates case-based and explanation-based reasoning paradigms when realizing the abductive subtask of the overall diagnostic task.     

Contextual Probability and Neighbourhood Counting

In this paper, we review the concept of contextual probability, the resulting notion of neighbourhood counting and the various specialisations of this notion which result in new functions for measuring similarity, such as all common subsequences. We also provide new results on the generalisation of the all common subsequences similarity. Contextual probability was originally proposed as an alternative way of reasoning. It was later found to be an alternative way of estimating probability, and it led to the introduction of the neighbourhood counting notion. This notion was then found to be a generic similarity metric that can be applied to different types of data.     

Contextual Text Mining Framework for Unstructured Textual Judicial Corpora through Ontologies

Digitalization has changed the way of information processing, and new techniques of legal data processing are evolving. Text mining helps to analyze and search different court cases available in the form of digital text documents to extract case reasoning and related data. This sort of case processing helps professionals and researchers to refer the previous case with more accuracy in reduced time. The rapid development of judicial ontologies seems to deliver interesting problem solving to legal knowledge formalization. Mining context information through ontologies from corpora is a challenging and interesting field. This research paper presents a three tier contextual text mining framework through ontologies for judicial corpora. This framework comprises on the judicial corpus, text mining processing resources and ontologies for mining contextual text from corpora to make text and data mining more reliable and fast. A top-down ontology construction approach has been adopted in this paper. The judicial corpus has been selected with a sufficient dataset to process and evaluate the results. The experimental results and evaluations show significant improvements in comparison with the available techniques.     

Abductive Knowledge Base Updates for Contextual Reasoning

We show in this paper how procedures that update knowledge bases can naturally be adapted to a number of problems related to contextual reasoning. The fact that the update procedures are abductive in nature is favourably exploited to tackle problems related to human-computer dialogue systems. We consider as examples aspects of pronoun resolution,goal formulation , and the problem of restoring the consistency of a knowledge base after some knowledge update is carried out. We state these problems in terms of the update problem and abductive reasoning and show how procedures that update knowledge bases yield some interesting results. We also explain how these procedures can naturally be used to model various forms of hypothetical reasoning such as hypothesizing inconsistencies and performing some look ahead form of reasoning.We do not claim thaT the problems presented here are solved entirely within the update framework. However, we believe that the flexibility of the representation and of the problem-solving approach suggest that the problems could be solved by adding more details about each problem. What is most interesting in our understanding is that all the aforementioned problems are expressed and tackled within the same framework.     

Neurocomputing hardware: present and future

This paper discusses the current state of the art of industrial neurocomputing, and then speculates on its future.Three examples of commercial neuro-silicon are presented: the Adaptive Solutions CNAPS system, the Intel ETANN chip, and the Synaptics OCR chip.We then speculate on where commercial neurocomputing hardware is going. In particular we propose that commercial systems will evolve in the direction of capturing more contextual, knowledge level information. Some results of an industrial handwritten character recognition system created at Apple Computers will be presented which demonstrate the power of adding contextual knowledge to neural network based recognition. Also discussed will be some of the possible directions required for neural network algorithms needed to capture such knowledge and utilize it effectively, as well as results from experiments on capturing contextual knowledge using several different neural network algorithms.Finally, the issues involved in designing VLSI architectures for the efficient emulation of sparsely activated, sparsely connected contextual networks will be discussed. There are fundamental cost/performance limits when emulating such sparse structures in both the digital and analog domain.     

一种上下文移动用户偏好自适应学习方法

针对移动网络对个性化移动网络服务系统的性能提出了更高的要求,但现有研究难以自适应地修改上下文移动用户偏好以为移动用户提供实时、准确的个性化移动网络服务的问题,提出了一种上下文移动用户偏好自适应学习方法,在保证精确度的基础上缩短了学习的响应时间.首先,通过分析移动用户行为日志来判断移动用户行为是否受上下文影响,并在此基础上判断移动用户行为是否发生变化.然后,根据判断结果对上下文移动用户偏好进行修正.在对发生变化的上下文移动用户偏好进行学习时,将上下文引入到最小二乘支持向量机中,进一步提出了基于上下文最小二乘支持向量机(C-LSSVM)的上下文移动用户偏好学习方法.最后,实验结果表明,当综合考虑精确度和响应时间两方面因素时,所提出的方法优于其他学习方法,并且可应用于个性化移动网络服务系统中.     

On Transitions Between Representations: The Role of Contextual Reasoning in Calculus Problem Solving

This article argues for a shift in how researchers discuss and examine students' uses and understandings of multiple representations within a calculus context. An extension of Zazkis, Dubinsky, and Dautermann's (1996) visualization/analysis framework to include contextual reasoning is proposed. Several examples that detail transitions between modes of reasoning and how these transitions inform students' reasoning in a calculus context are discussed. These examples are used to provide evidence for the usefulness of the model for unpacking student reasoning.     

Contextual reasoning distilled

In this paper we provide a foundation of a theory of contextual reasoning from the perspective of a theory of knowledge representation. Starting from the so-called metaphor of the box, we firstly show that the mechanisms of contextual reasoning proposed in the literature can be classified into three general forms (called localized reasoning, push and pop, and shifting). Secondly, we provide a justification of this classification, by showing that each mechanism corresponds to operating on a fundamental dimension along which context dependent representations may vary (namely, partiality, approximation and perspective). From the previous analysis, we distill two general principles of a logic of contextual reasoning. Finally, we show that these two principles can be adequately formalized in the framework of MultiContext Systems. In the last part of the paper, we provide a practical illustration of the ideas discussed in the paper by formalising a simple scenario, called the Magic Box problem.     

马尔可夫平稳随机域模型与遥感图象空间邻域结构信息分析

遥感图象记录的是景物对电磁波的反射与辐射能量及其空间分布信息。对一个象元的判别不仅取决于其光谱值,还取决于该象元的空间位置及与其它象元的关系,即邻域结构关系。本文主要讨论遥感图象空间邻域结构信息分析方法。首先,提出了三种图象平面点阵平稳马尔可夫模型,并论证了相应的计算方法;然后给出了模型中相关参数的最小二乘估计;基于给出的三种模型,我们采用联合概率密度函数作为准则函数,并利用递归分类方法逐步改善近邻类别知识,从而改善邻域结构分类结果。试验证明了所提出的三种模型的有效性。