基于改进多嵌入空间的实时语义数据流推理

将语义数据流处理引擎与知识图谱嵌入表示学习相结合,可以有效提高实时数据流推理查询性能,但是现有的知识表示学习模型更多关注静态知识图谱嵌入,忽略了知识图谱的动态特性,导致难以应用于实时动态语义数据流推理任务。为了使知识表示学习模型适应知识图谱的在线更新并能够应用于语义数据流引擎,建立一种基于改进多嵌入空间的动态知识图谱嵌入模型PUKALE。针对传递闭包等复杂推理场景,提出3种嵌入空间生成算法。为了在进行增量更新时更合理地选择嵌入空间,设计2种嵌入空间选择算法。基于上述算法实现PUKALE模型,并将其嵌入数据流推理引擎CSPARQL-engine中,以实现实时语义数据流推理查询。实验结果表明,与传统的CSPARQL和KALE推理相比,PUKALE模型的推理查询时间分别约降低85%和93%,其在支持动态图谱嵌入的同时能够提升实时语义数据流推理准确率。     

PPDM中面向k-匿名的MI Loss评估模型

隐私保护数据挖掘（PPDM）利用匿名化等方法使数据所有者在不泄露隐私信息的前提下,安全发布在数据挖掘中有效可用的数据集。k-匿名算法作为PPDM研究使用最广泛的算法之一,具有计算开销低、数据形变小、能抵御链接攻击等优点,但是在一些k-匿名算法研究中使用的数据可用性评估模型的权重设置不合理,导致算法选择的最优匿名数据集在后续的分类问题中分类准确率较低。提出一种使用互信息计算权重的互信息损失（MI Loss）评估模型。互信息反映变量间的关联关系,MI Loss评估模型根据准标识符和标签之间的互信息计算权重,并通过Loss公式得到各个准标识符的信息损失,将加权后的准标识符信息损失的和作为数据集的信息损失,以弥补评估模型的缺陷。实验结果证明,运用MI Loss评估模型指导k-匿名算法能够明显降低匿名数据集在后续分类中的可用性丢失,相较于Loss模型和Entropy Loss模型,该模型分类准确率提升了0.73%~3.00%。     

卫生经济学应用于慢性病防控决策的战略研究

为提高当前慢性病防控体系的效率,更好地遏制慢性病的流行,保护公众健康,2015年中国工程院设立了"卫生经济学应用于慢性病防控决策的战略研究"重大咨询项目。课题组调研发现,作为一种重要的卫生决策工具,卫生经济学在慢性病防控决策过程中尚处于初级应用阶段。对卫生经济学的重要性认识不足,对卫生经济学的方法的掌握和应用能力有限,制约了其在慢性病防控决策中的应用,因此建议建立多元主体参与的卫生经济学研究力量,加强卫生经济学应用于慢性病防控决策的大数据积累及应用研究,提出将卫生经济学应用于我国慢性病防控决策的战略框架。     

On mesh restrictions to satisfy comparison principles,maximum principles,and the non-negative constraint: Recent developments and new results

This article concerns mesh restrictions that are needed to satisfy several important mathematical properties—maximum principles, comparison principles, and the nonnegative constraint—for a general linear second-order elliptic partial differential equation. We critically review some recent developments in the field of discrete maximum principles, derive new results, and discuss some possible future research directions in this area. In particular, we derive restrictions for a three-node triangular (T3) element and a four-node quadrilateral (Q4) element to satisfy comparison principles, maximum principles, and the nonnegative constraint under the standard single-field Galerkin formulation. Analysis is restricted to uniformly elliptic linear differential operators in divergence form with Dirichlet boundary conditions specified on the entire boundary of the domain. Various versions of maximum principles and comparison principles are discussed in both continuous and discrete settings. In the literature, it is well-known that an acute-angled triangle is sufficient to satisfy the discrete weak maximum principle for pure isotropic diffusion. Herein, we show that this condition can be either too restrictive or not sufficient to satisfy various discrete principles when one considers anisotropic diffusivity, advection velocity field, or linear reaction coefficient. Subsequently, we derive appropriate restrictions on the mesh for simplicial (e.g., T3 element) and nonsimplicial (e.g., Q4 element) elements. Based on these conditions, an iterative algorithm is developed to construct simplicial meshes that preserve discrete maximum principles using existing open source mesh generators. Various numerical examples based on different types of triangulations are presented to show the pros and cons of placing restrictions on a computational mesh. We also quantify local and global mass conservation errors using representative numerical examples and illustrate the performance of metric.     

A Change-Point Approach for Phase-I Analysis in Multivariate Profile Monitoring and Diagnosis

Process monitoring and fault diagnosis using profile data remains an important and challenging problem in statistical process control (SPC). Although the analysis of profile data has been extensively studied in the SPC literature, the challenges associated with monitoring and diagnosis of multichannel (multiple) nonlinear profiles are yet to be addressed. Motivated by an application in multioperation forging processes, we propose a new modeling, monitoring, and diagnosis framework for phase-I analysis of multichannel profiles. The proposed framework is developed under the assumption that different profile channels have similar structure so that we can gain strength by borrowing information from all channels. The multidimensional functional principal component analysis is incorporated into change-point models to construct monitoring statistics. Simulation results show that the proposed approach has good performance in identifying change-points in various situations compared with some existing methods. The codes for implementing the proposed procedure are available in the supplementary material.     

基于自然冷却技术应用的数据中心空调节能分析

本文基于数据中心空调系统的选型和节能设计,以风冷直膨式机组、风冷双冷源冷水机组和热管复合式制冷机组为研究对象,建立了三种类型空调系统的能效计算模型;基于DEST提供的广州、上海、北京和哈尔滨四个城市的气象数据,以相同的空调负荷和回风参数为条件,对三类空调的年均能效(AEER)进行计算和对比分析。结果表明:三类机组的年均能效自南向北随着平均气温的降低而逐渐提高,分别为4.07~4.45、4.00~6.84、4.27~8.20;相比前二类机组,热管复合式制冷机组的节能幅度分别为4.8%~46%和6.4%~16.6%;双冷源制冷机组由于能够应用自然冷却替代或部分替代高能耗的压缩制冷,相比风冷直膨式机组节能优势明显;与风冷双冷源冷水机组相比,热管复合式制冷机组的制冷剂输送功小、换热端工作效率高、没有防冻问题等优点,是年均气温较低地区数据中心空调的首选。     

Enhanced assessment of ageing phenomena in steel structures based on materials data and non‐destructive testing

The increasing amount of ageing civil steel infrastructure requests an enhanced assessment of this infrastructure in terms of determining its residual fatigue life in a more realistic way than this has been done in the past. Often the relevant materials data for cyclic loading of such an ageing infrastructure is not available and its retrieval turns out to be relatively cumbersome bearing the urgency in data availability and continuous cost pressure in mind. This article addresses different approaches and techniques on how materials data for cyclic loading can be obtained at a fraction of the effort compared to state‐of‐the‐art techniques, considering load increase tests, non‐destructive testing techniques and finally even a stepped bar specimen allowing a complete set of materials data (stress‐strain behaviour and stress‐ and strain‐life curve) to be obtained with a single specimen in the end only. Options for ’digitizing’ materials data evaluation are discussed and some prospect on application of those novel approaches and techniques in damage accumulation assessments on real steel infrastructure is provided.     

关于APP跨类别交叉推广的精准营销小数据设计研究

目的挖掘大学生群体的潜在需求,辅助企业进行交叉推广的精准营销,在信息筛选上以更有针对性的方式设计移动端产品。方法收集大学生群体使用手机超过6600个APP的小数据,运用统计学相关性原理对跨类别应用进行相关性分析。结论小数据分析发现大学生群体对社交类的兴趣偏好最强,其中女性更关注摄影与录像类。相关性方面,生活类、旅游类、美食佳饮类三者之间的相关性最高。交叉推广的方式上,健康健美类与导航类可以依附于摄影与录像类,音乐类和效率类可以依附于社交类进行辅助推广,而摄影与录像类和生活类可以相互推广,男性群体中的参考类和女性群体中的新闻类,均可和商务类以及图书类相互推广。可将兴趣偏好、相关性以及兴趣偏差纳入大学生群体的交叉推广中,进行信息筛选和设计。