分布式策略与CRFs相结合识别汉语组块

该文提出了一种基于CRFs的分布式策略及错误驱动的方法识别汉语组块。该方法首先将11种类型的汉语组块进行分组,结合CRFs构建不同的组块识别模型来识别组块;之后利用基于CRFs的错误驱动技术自动对分组组块进行二次识别;最后依据各分组F值大小顺序处理类型冲突。实验结果表明,基于CRFs的分布式策略及错误驱动方法识别汉语组块是有效的,系统开放式测试的精确率、召回率、F值分别达到94.90%、91.00%和92.91%,好于单独的CRFs方法、分布式策略方法及其他组合方法。     

VN结构识别的一种概率分布模型

正确识别汉语里的VN结构等基本名词短语可以帮助提高句法分析的准确率。提出并验证了如果动名组合的上下文词语的分布类似,那么它们的结构也类似的假设。结合动词、名词本身,构造了一种基于概率分布的结构向量空间模型,用于VN结构的识别。实验结果表明,虽然没有使用其他外部资源,该方法仍取得了理想的识别效果,精确率和召回率分别达到了95.2%和93.0%。     

语义信息与CRF结合的汉语功能块自动识别

为了构建汉语功能块自动识别系统,该文利用条件随机域模型对经过正确词语切分和词性标注处理的汉语句子进行功能块边界识别和功能信息标注处理,通过在特征提取阶段优化组合丰富的上下文特征,得到功能块识别的精确率、召回率和F1-measure值分别为85.84%、85.07%和85.45%。在此基础上,该文引入由词义聚合关系将汉语单词组织起来的《同义词词林》作为语义资源,把其中的语义信息作为特征加入到功能块的识别过程,缓解了数据稀疏以及歧义问题对识别结果造成的影响,使得上述三个性能指标分别提高到86.21%、85.31%和85.76%。     

汉语最长名词短语的自动识别

通过对包含5573个汉语句子的语料文本中的最长名词短语的分布特点的统计分析,提出了两种有效的汉语最长名词短语自动识别算法：基于边界分布概率的识别算法和基于内部结构组合的识别算法.实验结果显示,后者的识别正确率和召回率分别达到了85.4%和82.3%,取得了较好的自动识别效果.     

限定领域的汉语-维吾尔语句子级对齐研究

针对政府文献的汉语维吾尔语语料库特点,充分利用汉语和维吾尔语的句子特性,提出一种汉维句子级别的对齐方法。该方法重点分析政府领域的汉语和维吾尔语的句型,分别对汉语和维吾尔语的语料进行边界识别,避免了复杂句型对汉语-维吾尔语句子对齐的影响,使得该方法取得句子对齐达到97%与99%之间的正确率。对齐的汉语-维吾尔语句子对可以充实语料库的规模,为汉语-维吾尔语短语对齐以及汉维机器翻译提供翻译语料。     

一种对汉语孤立字词进行声韵分割的时频方法

正确的声韵分离是汉语语音识别与合成等的基础和关键。针对传统声韵分割中,时域短时能量和过零率容易受到噪声干扰从而导致分割不准确的问题,结合语谱图所体现的时频信息对汉语孤立字进行了声韵分割,并进一步对信号进行经验模态分解和计算保号率,实现了一种对二字词的时频声韵分割方法。仿真实验结果显示,该方法对汉语孤立字和二字词的分割准确率分别达到了86.92%和77.47%。     

以“的”字结构为核心的最长名词短语识别研究

以“的”字结构为核心的最长名词短语是汉语最长名词短语的一个特殊子类。以该短语的自动识别为基础重新分化了汉语MNP的识别任务。在考察其结构和分布特征的基础上,提出“先识别右边界,识别成果参与左边界识别”的策略,并使用边界分布概率模型分治了左右边界。实验基于85万字的新闻语料上进行训练,并在42万字的同质语料上进行了开放测试,取得了80.63%的正确率和75.68%的召回率。     

基于SVM融合多特征的介词结构自动识别

介词结构在汉语文本中出现频率很高,正确识别介词结构边界对句法分析、语音合成中的韵律短语划分有着重要意义。该文较为系统地探讨了汉语中常用介词的边界识别问题。利用支持向量机SVM模型,基于输出概率而不是简单的二分法来选择正确的后边界。探讨了不同的特征选择,并尝试加入语义信息等不同特征组合以提高识别准确率。对常用的68个介词进行边界识别实验,5折交叉验证的准确率达到90.95%,优于前人的识别结果。     

基于遗传算法的汉语未登录词识别

针对汉语分词处理中未登录词识别这一难点,提出一种应用遗传算法识别的新方法.该方法扩大了分词碎片容量,将未登录词识别问题看成二值分类问题,即在预处理后产生的分词碎片中,单字存在"可组合"和"不可组合"两类,采用遗传算法先将分词碎片中的单字词确定下来,然后将其余相邻单字组合,完成未登录词识别.实验结果表明,该方法可有效地解决未登录词识别问题,提高未登录词识别的精确率和召回率.     

混合的汉语基本名词短语识别方法

提出一种混合的汉语基本名词短语（BaseNP）识别模型,包括采用语法规则、统计方法和组合分类器方法。利用BaseNP词的信息、词性信息及上下文句法信息,构建组合分类器,提高判断的准确性。在中文树库（CTB5.0）上进行实验,F值达到了90.09%,证明该方法能有效地识别BaseNP。     

Efficient Online Virtual Network Mapping Using Resource Evaluation

Network virtualization is a promising solution that can prevent network ossification by allowing multiple heterogeneous virtual networks (VNs) to cohabit on a shared substrate network. It provides flexibility and promotes diversity. A key issue that needs to be addressed in network virtualization is allocation of substrate resources for the VNs with respect to their resource requirements and the topologies of the substrate and virtual networks, namely the VN mapping (VNM) problem. Efficient VNM algorithms aim to maximize the number of coexisting VNs, and increase the utilization and revenue obtained from the substrate resources. In this paper, we present an online VNM algorithm (OVNM) that maximizes the number of coexisting VNs leading to good utilization and revenue of the substrate. Using the OVNM algorithm, we estimate the VN mapping and evaluate the associated substrate resources to map the VN within a proper region on the substrate by using the FVN_Sort (first virtual node sorting) function. This improves the probability of a VN mapping success. Furthermore, by mapping the virtual nodes and links in a coordinated fashion, the resource consumption while mapping is minimized. We evaluate the performance of our approach by using simulation, and show that the algorithm has an acceptable run time and leads to a better blocking probability performance, which means more coexisting VNs.     

基于知识描述和遗传算法的跨域虚拟网络映射

网络虚拟化环境下的跨域虚拟网络映射是指当物理网络由多个自治域构成时,以最小化虚拟网络映射开销为目标,将虚拟网络请求恰当地划分为多个虚拟子网请求,并分别指派给相应自治域以完成映射。资源匹配和虚拟网络划分是跨域虚拟网络映射中的两个关键阶段。然而,现有的资源匹配算法无法支持精确的数值属性匹配,也无法满足虚拟网络用户对表达多样化映射约束的需求,故实用性不高。此外,虚拟网络划分属于NP问题,目前也缺乏高效的求解方法。针对上述两个阶段中存在的问题,分别提出了基于OWL及SWRL的资源匹配算法和基于遗传算法的虚拟网络划分算法。理论分析证明了该方法的正确性。仿真实验从效率、性能及稳定性方面验证了该方法的有效性。     

Optimal virtual network embedding: Energy aware formulation

Network Virtualization is a key component of the Future Internet, providing the dynamic support of different networks with different paradigms and mechanisms in the same physical infrastructure. A major challenge in the dynamic provision of virtual networks is the embedding approach taking energy efficiency into account, while not affecting the overall Virtual Network (VN) acceptance ratio. Previous research focused on either designing heuristic-based algorithms to address the efficient embedding problem or to address the energy impact.This paper proposes an integer linear programming formulation, Energy Aware–Virtual Network Embedding–Node-Link Formulation (EA–VNE–NLF), that solves the online virtual network embedding as an optimization problem, striving for the minimum energy consumption and optimal resource allocation per VN mapping. Two different objective functions are proposed: (i) addressing primarily the resource consumption problem – Bandwidth Consumption Minimization (BCM); (ii) addressing primarily the energy consumption problem – Energy Consumption Minimization (ECM).The performance of each objective function is evaluated by means of simulation and compared with an existing objective function, Weighted Shortest Distance Path (WSDP), that is considered state of the art of the resource allocation problem. The simulation results show that the objective function BCM reduces the energy consumption of the physical network by 14.4%, and improves the embedding factor by 4.3%, consuming almost the same amount of resources as requested, and slightly worsening the VN acceptance ratio by 2.3%. ECM reduces the energy consumption of the physical network by 31.4% and improves the embedding factor by 4.1%, without affecting the VN acceptance ratio when compared to WSDP.     

A virtual network mapping algorithm based on integer programming

The virtual network （VN） embedding/mapping problem is recognized as an essential question of network virtualization. The VN embedding problem is a major challenge in this field. Its target is to efficiently map the virtual nodes and virtual links onto the substrate network resources. Previous research focused on designing heuristic-based algorithms or attempting two-stage solutions by solving node mapping in the first stage and link mapping in the second stage. In this study, we propose a new VN embedding algorithm based on integer programming. We build a model of an augmented substrate graph, and formulate the VN embedding problem as an integer program with an objective function and some constraints. A factor of topology-awareness is added to the objective function. The VN embedding problem is solved in one stage. Simulation results show that our algorithm greatly enhances the acceptance ratio, and increases the revenue/cost （R/C） ratio and the revenue while decreasing the cost of the VN embedding problem.     

Virtual network embedding for power savings of servers and switches in elastic data center networks弹性数据中心网络中的服务器和交换机联合节能虚拟网络嵌入机制

Currently, the elastic interconnection has realized the high-rate data transmission among data centers (DCs). Thus, the elastic data center network (EDCN) emerged. In EDCNs, it is essential to achieve the virtual network (VN) embedding, which includes two main components: VM (virtual machine) mapping and VL (virtual link) mapping. In VM mapping, we allocate appropriate servers to hold VMs. While for VL mapping, an optimal substrate path is determined for each virtual lightpath. For the VN embedding in EDCNs, the power efficiency is a significant concern, and some solutions were proposed through sleeping light-duty servers. However, the increasing communication traffic between VMs leads to a serious energy dissipation problem, since it also consumes a great amount of energy on switches even utilizing the energy-efficient optical transmission technique. In this paper, considering load balancing and power-efficient VN embedding, we formulate the problem and design a novel heuristic for EDCNs, with the objective to achieve the power savings of servers and switches. In our solution, VMs are mapped into a single DC or multiple DCs with the short distance between each other, and the servers in the same cluster or adjacent clusters are preferred to hold VMs. Such that, a large amount of servers and switches will become vacant and can go into sleep mode. Simulation results demonstrate that our method performs well in terms of power savings and load balancing. Compared with benchmarks, the improvement ratio of power efficiency is 5%–13%.     

基于二级控制器和线性模型的自适应带宽分配算法

针对虚拟化网络中存在的带宽分配问题,提出了一种基于二级控制器系统的自适应带宽分配算法。该控制器系统由包含VN子控制器的一组SP控制器和InP控制器构成;首先把每个VN在每个控制间隔的特性用一个线性模型来局部近似,以得到其过去、当前的带宽分配以及过去包时延性能之间的关系,然后基于物理链路容量的限制,系统对每个VN所请求的带宽作出响应并进行动态调整,以达到灵活的分布式自治控制,并保证为每个VN分配合理的带宽。仿真结果表明,提出的算法不仅能够在多个VN之间合理、动态地分配可用链路容量,而且能够提高总带宽的利用率和分配的公平性,满足每个VN的QoS需求。     

基于柴油机虚拟样机的动力学仿真研究

通过对柴油机部件装配进行机构运动学和动力学仿真，能为柴油机的结构和性能优化提供依据。该文基于虚拟样机技术及其支撵软件VN4D和Pro／E，对12VPA6柴油机的曲轴-连杆-活塞机构进行了运动学、动力学仿真研究。首先，运用三维建模软件Pro／E，建立了12VPA6型柴油机曲轴-连杆-活塞机构的虚拟样机模型，然后将模型导入仿真软件VN4D中，对曲轴-连杆-活塞运动件进行了运动学和动力学仿真。通过计算机对柴油机虚拟样机的仿真运行，获取了仿真模型的运动学、动力学特性数据，得到了与理论分析相吻合的结果，为进一步对该型柴油机曲轴-连杆-活塞机构进行有限元分析奠定了基础。     

A graphical approach for the evaluation of land-cover classification procedures

Abstract

A method for evaluating the effectiveness of different feature combinations and training strategies is described. Preliminary tests have been made using two groups of feature combinations derived from SPOT High Resolution Visible (HRV) data and two sets of training samples. The method is objective, and needs no ground confirmation or interaction from the image analyst. It is recommended as a surrogate for detailed accuracy assessment when attempting to find an optimum set of training pixels or feature combinations for image classification.