基于鱼群算法的异构数据库语义聚类的研究

数据库语义上的异构成为数据整合的重点和难点。针对该问题文章提出一种基于鱼群算法的异构数据库语义聚类算法。其思路是,首先对数据库属性信息进行向量化、矢量化,利用鱼群算法得到一个较优的聚类结果,再根据鱼群算法的结果使用模糊C-均值算法进行调整聚类。实例分析结果证明,该算法具有较高的聚类准确度。     

An automatic subdigraph renovation plan for failure recovery of composite semantic Web services

A Web service-based system never fulfills a user＇s goal unless a failure recovery approach exists. It is inevitable that several Web services may either perish or fail before or during transactions. The completion of a composite process relies on the smooth execution of all constituent Web services. A mediator acts as an intermediary between providers and consumers to monitor the execution of these services. If a service fails, the mediator has to recover the whole composite process or else jeopardize achieving the intended goals. The atomic replacement of a perished Web service usually does not apply because the process of locating a matched Web service is unreliable. Even the system cannot depend on the replacement of the dead service with a com- posite service. In this paper, we propose an automatic renova- tion plan for failure recovery of composite semantic services based on an approach of subdigraph replacement. A replacement subdigraph is posed in lieu of an original subdigraph, which includes the failed service. The replacement is done in two separate phases, ofltine and online, to make the recovery faster. The ofitine phase foresees all possible subdigraphs, pre-calculates them, and ranks several possible replacements. The online phase compensates the unwanted effects and executes the replacement subdigraph in lieu of the original subdigraph. We have evaluated our approach during an experiment and have found that we could recover more than half of the simulated failures. These achievements show a significant improvement compared to current approaches     

Algorithms for checking channel passing in web service choreography

Web service choreography describes global mod- els of service interactions among a set of participants. For an interaction to be executed, the participants must know the required channel（s） used in the interaction, otherwise the ex- ecution will get stuck. Since channels are composed dynami- cally, the initial channel set of each participant is often insuf- ficient to meet the requirements. It is the responsibility of the participants to pass required channels owned （known） by one to others. Since service choreography may involve many par- ticipants and complex channel constraints, it is hard for de- signers to specify channel passing in a choreography exactly as required. We address the problem of checking whether a service choreography lacks channels or has redundant chan- nels, and how to automatically generate channel passing based on interaction flows of the service choreography in the case of channel absence. Concretely, we propose a sim- ple language Chorc, a channel interaction sub-language for modeling the channel passing aspect of service choreography. Based on the formal operational semantics of Chore, the algo- rithms for static checking of service choreography and gen- erating channel passing are also studied, and the complexity results of algorithms are discussed. Moreover, some illus- trated service choreography examples are presented to show how to formalize and analyze service choreography with channel passing in Chorc.     

文本语义分析的实现及应用

淘宝网每天产生千万级的用户输入文本（简称UGC）,它们是无结构的,因此与结构化数据相比,更难通过语义分析技术进行信息挖掘。然而,UGC的无结构化,又使得其能够挖掘的信息是无限的。本文将讲述如何利用文本语义分析技术来挖掘有价值的信息,并介绍该技术在淘宝网的应用情况。     

一种面向语义Web服务的语义程序变换方法

语义程序变换是面向语义Web服务的软件设计方法的基础,语义程序只有通过程序变换后才能被运行环境执行和调用,然而目前还缺乏有效的语义程序变换方法。针对这一问题,基于语义编程语言SPL,提出了一种面向语义Web服务的语义程序变换方法。该方法通过对语义数据类型、语义规则、语义服务和语义流程等语义信息的有效变换,不仅提高了面向服务的程序设计的灵活性和健壮性,而且有助于提高业务流程的柔性和重用性。     

面向最终用户编程的服务虚拟化模型

为了解决面向最终用户编程问题,提出了一种基于动态语义的服务虚拟化模型。该模型以SOA思想理论为基础,采用面向服务计算和Web服务技术,结合本体服务描述语言和描述逻辑,建立服务虚拟化模型,从而实现面向最终用户编程。该模型有效地提高了模型的查全率和查准率,能够在一定程度上解决软件危机。     

可信计算语义证明模型的研究

针对现有的可信计算平台体系认证技术是静态的、一次性的,并不基于语义等问题,提出了语义远程认证抽象模型,该模型由代码特征化部件、安全监测部件、安全控制部件、应用接口部件、知识库组成。其中关键问题知识库采用离线创建规则的学习模式,供在线检测。     

游戏中图像内容情感语义的分析与研究

针对目前游戏中忽略情感因素的问题,在游戏场景中引入情感语义的概念,提出基于情感语义的游戏场景图像检索系统和情感计算研究情感信息的数字化处理方法,采用支持向量机算法建立场景图像内容和其所表达的情感语义之间的联系。来建立一个能感知识别人的情感,能对人的情感做出智能反应的游戏系统,从而使人机交互更加和谐化。     

图像区域粗糙分割情况下的区域物体分类

提出一种改进的语义物体分类方法,可在图像区域分割结果不精确的情况下,对图像区域进行分类.结合统计文本分析中的bag-of-words方法与多示例学习,将图像区域物体切分为若干小图像块,提取图像区域物体的图像特征作为其粗糙语义概念并计算置信度;根据粗糙语义概念进一步提取出各种区域物体类型的语义特征作为其特征语义概念;使用分类器对特征语义概念进行学习,实现了对区域物体的分类.实验结果表明,采用文中方法可在分割粗糙的图像区域上获得很好的区域物体分类准确率.     

集成模糊LSA与MIL的图像分类算法

针对自然图像的分类问题,提出一种基于模糊潜在语义分析(LSA)与直推式支持向量机(TSVM)相结合的半监督多示例学习(MIL)算法.该算法将图像当作多示例包,分割区域的底层视觉特征当作包中的示例.为了将MIL问题转化成单示例问题进行求解,首先利用K-Means方法对训练包中所有的示例进行聚类,建立"视觉词汇表";然后根据"视觉字"与示例之间的距离定义模糊隶属度函数,建立模糊"词-文档"矩阵,再采用LSA方法获得多示例包(图像)的模糊潜在语义模型,并通过该模型将每个多示例包转化成单个样本;采用半监督的TSVM训练分类器,以利用未标注图像来提高分类精度.基于Corel图像库的对比实验结果表明,与传统的LSA方法相比,模糊LSA的分类准确率提高了5.6%,且性能优于其他分类方法.