一种优化的基于LPI的文档索引算法FLPI

LPI对于局部流形结构是优化的, 但在时空上运行效率较低,使其很难应用于大型数据集。基于LPI算法,提出了一种优化的LPI算法FLPI,它将LPI问题分解为一个图嵌入问题和一个正则最小二乘问题,避免了稠密矩阵的特征值分解,显著减少了计算复杂度。此外,在监督环境下,利用一个特别设计的图,使FLPI只需要解决正则最小二乘问题,进一步减少了时空开销。实时数据集实验结果显示,FLPI获得了相似或优于LPI的结果,且运行速度明显提升。     

基于非负矩阵分解的中文文本主题分类

提出基于非负矩阵分解（NMF）的中文文本主题分类方法,应用NMF算法分解词．文本矩阵获取词之间的相关性,有效地解决同义词、多义词的影响。实验结果表明,与基于奇异值分解的潜在语义索引方法相比,该方法计算速度快、占用存储空间较少。在潜在语义数据降低较大的情况下,NMF方法具有更好的分类精度。     

一种大规模数据的快速潜在语义索引

潜在语义索引（LSI）已应用到现代信息检索的多个领域,但矩阵奇异值分解的高复杂度阻碍了该技术在大规模数据上的应用。提出一种大规模数据的快速LSI方法。给出一个降维问题的统一框架,LSI作为一种特征提取算法,可以在这个框架下转化为一个特征选择问题。利用该技术在最大程度保持LSI降维效果的同时,简化LSI的计算,使其能够应用于大规模数据。     

使用基于SVM的局部潜在语义索引进行文本分类

潜在语义索引(LSI)通过奇异值分解(SVD)获得原始词—文档矩阵的潜在语义结构,在一定程度上解决了一词多义和多词一义问题。但目前文本分类中使用LSI方法的效果并不理想,这是因为没有充分考虑分类信息。为解决该问题,提出一种改进的局部潜在语义索引(LLSI)方法,使用支持向量机(SVM)来产生局部区域。实验结果表明,该方法是有效的。     

Improving customer complaint management by automatic email classification using linguistic style features as predictors

Customer complaint management is becoming a critical key success factor in today's business environment. This study introduces a methodology to improve complaint-handling strategies through an automatic email-classification system that distinguishes complaints from non-complaints. As such, complaint handling becomes less time-consuming and more successful. The classification system combines traditional text information with new information about the linguistic style of an email. The empirical results show that adding linguistic style information into a classification model with conventional text-classification variables results in a significant increase in predictive performance. In addition, this study reveals linguistic style differences between complaint emails and others.     

基于LSI和SVM分类法的定题邮件过滤研究

潜在语义索引(LSI)是一种有效的信息查询方法,同时也被成功地应用到了文本分类中。LSI能解决同义和多义的问题,通过降低原始文档-术语矩阵的噪声来凸现出词条和文档之间的语义关系。为了识别和过滤有害的、不期望的定题的信息或Email,在双语言环境下(包括中文和英文),提出了一个基于改进的LSI方法的定题邮件类信息过滤系统,该系统采用潜在语义模型来表示被过滤的信息类,通过奇异值分解和正例监护学习方法,选择支持向量机(SVM)来识别和分类预定义的定题信息。实验结果表明:基于LSI的特征选择的SVM分类算法是一种更有效的信息识别和文本分类方法,不但具有较好的分类性能,同时也能大大减小计算的复杂性。     

基于支持向量机的隐含语意特征选择方法

隐含语意索引(LSI)是一个能有效捕获文档中词的隐含语意特征的方法。然而,用该方法选择的特征空间对文本分类来说可能不是最适合的,因为这种方法按照词的变化排序特征,而没有考虑到分类能力。支持向量机(SVM)高度的泛化能力使它特别适用于高维数据例如文档的分类。为此提出基于支持向量机的特征提取方法用于选择适于分类的LSI特征。该方法利用SVM高度泛化的分类能力, 通过使用在每一个规则下训练的分类器的参数对第k个特征对反向平方分解面的贡献w2k的值进行估计。实验表明当需要比LSI更少的训练和测试时间时,该方法能够以更为紧凑的表示方式提高分类性能。     

一种大容量文本集的智能检索方法

分析了潜在语义模型,研究了潜在语义空间中文本的表示方法,提出了一种大容量文本集的检索策略。检索过程由粗粒度非相关剔除和相关文本的精确检索两个步骤组成。使用潜在语义空间模型对文本集进行初步的筛选,剔除非相关文本;使用大规模文本检索方法对相关文本在段落一级进行精确检索,其中为了提高检索的执行效率,在检索算法中引入了遗传算法;输出这些候选的段落序号。实验结果证明了这种方法的有效性和高效性。     

GrOWL: A tool for visualization and editing of OWL ontologies

In an effort to optimize visualization and editing of OWL ontologies we have developed GrOWL—a browser and visual editor for OWL that accurately visualizes the underlying DL semantics of OWL ontologies while avoiding the difficulties of the verbose OWL syntax. In this paper, we discuss GrOWL visualization model and the essential visualization techniques implemented in GrOWL.     

Evaluating ontologies: Towards a cognitive measure of quality

Ontologies are formal specifications of shared conceptualizations of a domain. Important applications of ontologies include distributed knowledge-based systems, such as the semantic web, and the evaluation of modelling languages, e.g. for business process or conceptual modelling. These applications require formal ontologies of good quality. The quality of a formal ontology requires both a good conceptualization of a domain and a good specification of the conceptualization. In this paper, we focus on the latter aspect, and present a method to test how well a specification of a formal ontology corresponds to a conceptualization of a domain held by ontology users. Our experimental method is based on principles of cognitive psychology. We present two experiments to demonstrate our method using upper-level ontologies.     

基于主题词对的文档重排方法

信息检索中相关文档的排序一直是一个至关重要的问题。本文提出一种基于主题词对的文档重排方法,使得检索结果在保持召回率的前提下提高精确率。主题词对意指能够共同表征同一主题的两个词语,其中一个来自于查询,另一个来自于文档,两者之间具有紧密的联系。本文中,主题词对的选择采用概率潜在语义索引的方法,并根据主题词对在文档中的分布状况对其进行重排。对NTCIR-5中文信息检索的文档集合进行测试,采用trec标准评估方法,结果表明采用该方法使得精确率在rigid和relax结果集上分别提高了53.6% 和55.8%。     

Virtual organization security policies: An ontology-based integration approach

This paper addresses the specification of a security policy ontology framework to mediate security policies between virtual organizations (VO) and real organizations (RO). The goal is to develop a common domain model for security policy via semantic mapping. This mitigates interoperability problems that exist due to heterogeneity in security policy data among various (VO) and (RO) in the semantic web. We propose to carry out integration or mapping for only one aspect of security policy, which is authorization policy. Other aspects such as integrity, repudiation and confidentiality will be addressed in future work. We employ various tools such as Protégé, RacerPro and PROMPT to show proof of concept.

基于Boosting算法集成遗传模糊分类器的文本分类

提出一种新颖的基于Boosting模糊分类的文本分类方法。首先采用潜在语义索引(LSI)对文本特征进行选择;然后提出Boosting算法集成模糊分类器学习,在每轮迭代训练过程中,算法通过调整训练样本的分布,利用遗传算法产生分类规则。减少分类规则能够正确分类样本的权值,使得新产生的分类规则重点考虑难于分类的样本。实验结果表明,该文本分类算法具有良好分类的性能。     

Map merging for multiple robots using Hough peak matching

Navigation in a GPS-denied environment is an essential requirement for increased robotics autonomy. While this is in some sense solved for a single robot, the next challenge is to design algorithms for a team of robots to be able to map and navigate efficiently.The key requirement for achieving this team autonomy is to provide the robots with a collaborative ability to accurately map an environment. This problem is referred to as cooperative simultaneous localization and mapping (SLAM). In this research, the mapping process is extended to multiple robots with a novel occupancy grid map fusion algorithm. Map fusion is achieved by transforming individual maps into the Hough space where they are represented in an abstract form. Properties of the Hough transform are used to find the common regions in the maps, which are then used to calculate the unknown transformation between the maps.Results are shown from tests performed on benchmark datasets and real-world experiments with multiple robotic platforms.     

潜在语义分析理论及其应用

潜在语义分析(Latent Sereantic AnMysis,LSA)是一种用于自动地实现知识提取和表示的理论和方法,它通过对大量的文本集进行统计分析,从中提取出词语的上下文使用含义。在技术上,它同向量空间模型类型类似,都是采用空间向量表示文本,但通过SVD分解等处理,消除了同义词、多义词的影响,提高了后续处理的精度。将着重介绍LSA方法的基本思想、特点、实现方法,以及基于LSA思想的具体应用。     

基于事件的局部行为模型的合并

采用模态迁移系统描述系统行为,针对局部行为模型中存在的不确定行为,提出一种基于事件的局部行为模型合并方法。该方法首先定义局部行为模型之间的精化关系,利用精化关系产生合并规则,运用合并规则产生行为模型的极小共同精化模型或最小共同精化模型,从而消除局部行为模型中存在的不确定行为。最后通过一个示例对该方法的有效性作出说明。     

改进的概率潜在语义分析下的文本聚类算法

概率潜在语义分析(PLSA)模型用期望最大化(EM)算法进行参数训练,由于算法参数的随机初始化,致使聚类的效果过度拟合且过分依赖于参数初始值。将潜在语义分析(LSA)模型参数概率化,用以初始化概率潜在语义分析模型的参数,得到的改进算法有效解决了参数随机初始化问题。经实验验证,所提出的方法对文本聚类的归一化互信息(NMI)和准确度都有明显提高。     

Semantic Ontology Mapping for Interoperability of Learning Resource Systems using a rule-based reasoning approach

There is an increasing demand for sharing learning resources between existing learning resource systems to support reusability, exchangeability, and adaptability. The learning resources need to be annotated with ontologies into learning objects that use different metadata standards. These ontologies have introduced the problems of semantic and structural heterogeneity. This research proposes a Semantic Ontology Mapping for Interoperability of Learning Resource Systems. To enable semantic ontology mapping, this research proposes conflict detection and resolution techniques for both semantic and structural conflicts. The Semantic Bridge Ontology has been proposed as a core component for generating mapping rules to reconcile terms defined in local ontologies into terms defined in the target common ontology. This work defines the reasoning rules to classify related learning objects to enhance the powerful deductive reasoning capabilities of the system. As a consequence, ontology-based learning object metadata are generated and used by the semantic query engine to facilitate user queries of learning objects across heterogeneous learning resource systems.