采用数据血缘的数据热度预测方法

数据之间存在相互引用关系,在进行数据开发时,通常存在一些具有高热度的数据,此类数据被其他数据大量引用,它们的缺陷往往会给整个大数据平台产出的数据结果带来极大影响。因此,对高热度数据进行预测并予以相应保护至关重要。面向基于数据热度的数据分级治理需求,提出一种采用数据血缘的数据热度预测方法。首先通过构建数据系统中的数据血缘捕获数据节点之间的引用关系;然后,提取数据血缘的时间和结构特征,并采用图卷积网络(GCN)进行数据血缘图特征的学习;最后,提出一种数据血缘传播趋势分层读出的方法读出图特征,对数据热度进行预测。在浙江中烟营销系统数据集ZJZY-SL和高能物理现象学相关论文引文数据集(HEP-PH)上的实验结果表明,相较于DeepCCP等方法,所提方法的识别准确率分别提升7.64、2.88个百分点,平均F1分别提升4.7、4.34个百分点。所提方法能充分挖掘数据在被引用早期的数据血缘特征,并预测数据节点未来的热度。

命名数据网络中基于分级的数据缓存方法

针对NDN固有数据缓存机制存在数据内容多样性低、频繁替换等问题,提出了一种数据分级缓存方法.该方法将传输路径上的路由器进行等级划分,根据数据内容热度,将其缓存在相应等级的路由器中,以减少用户获取数据内容的时间.实验结果表明:该方法能够有效提高缓存命中率,减少平均路由跳数,降低平均请求时延.

微博热度数据采集系统研究与实现

为更好地开展微博热门话题监测工作,文中借鉴Web自动化功能测试技术,设计了一套基于WatirWebDriver的微博热度数据采集系统,实现了对微博及其热度数据的采集和处理。实验表明,系统能有效的解决传统采集方案中微博热度数据分散的问题,减轻后续分析工作压力并为其提供高质量数据素材。

一种基于数据热度的缓存策略

缓存是存储系统的关键部件之一,良好的缓存策略能够有效提升数据访问性能。传统基于时间局部性和访问频率的缓存策略已难以适应大数据背景下热点数据集中访问的需求。笔者提出一种基于数据热度的缓存策略(Hot Data Cached Based on Temperature,HDCT),综合考虑访问时间局部性和访问频率动态性,通过访问加热、周期性冷却动态调节数据热度,实现基于数据热度的缓存管理。实验表明,相对传统的LRU和LFU策略,HDCT策略在提升缓存命中率方面具有良好的效果。

基于SSD的机群文件系统元数据存储系统

随着数据量的增加和元数据操作性能需求的提高,传统基于磁盘(HDD)存储架构的机群文件系统元数据存储系统由于HDD成为性能瓶颈而无法满足需求.将SSD应用到元数据存储中,设计实现了一个基于SSD+HDD的异构元数据存储系统Hybrid MDSL.针对SSD的I/O特性设计了基于追加写的元数据存储组织,并通过基于访问热度的数据迁移机制提高SSD空间利用率.测试结果表明,Hybrid MDSL明显提高了元数据I/O的性能.

海量遥感数据的存储迁移策略研究

随着遥感技术的发展,遥感卫星数量增加、种类日趋丰富、分辨率不断提高,导致遥感数据量急速增长,给数据的存储管理效率提出了更高的要求。传统方法通常采用规则的方式对数据价值进行评估并形成相应的存储迁移策略,具有较强的主观因素,并未完全从数据的角度出发分析产品热度。从数据的访问规律着手,将数据分为访问频率高的热数据,访问频率相对较低的温数据和冷数据,利用机器学习算法对数据热度进行建模,结合磁盘阵列与蓝光光盘库等硬件存储设施特点,设计并实现一套基于蓝光光盘库与SAN存储的海量遥感数据的分级存储系统,并通过优化数据的迁移存储策略,实现海量遥感数据的高效、安全、节能存储。在实验平台上就2018年以来的产品需求进行了实验验证,表明采用提出的存储迁移策略比现有策略下系统服务请求的数据在线率提高44.4%。

基于时标分层技术的时序数据存储引擎研究与实现

针对使用时序数据库存储海量时序数据场景下,数据量增大导致数据写入和查询效率降低的问题,提出一种基于时标分层的时序数据存储引擎。通过将内存(RAM)、固态硬盘(SSD)、机械硬盘(HDD)三种存储介质存储特性与时序数据业务场景相结合,基于数据时标实现数据热度划分、数据分层与数据迁移操作,提高RAM和SSD使用率,提升数据写入和查询效率。结合该引擎对现有"海迅"实时库(HS)进行优化改造,改造后的数据库写入和查询性能分别提升70%和200%。

基于数据更新间隔的NAND闪存垃圾回收算法

闪存因具有速度快、体积小等优点而广泛应用于数据存储领域,为提高NAND闪存的垃圾回收效率、延长闪存使用寿命,提出一种基于数据更新间隔的垃圾回收算法UIGC。计算闪存中空闲页的分散度,将其作为垃圾回收触发条件。从垃圾回收效率和磨损均衡效果2个方面出发,综合考虑块中无效页年龄累计和以及块中有效页比例,使用动态回收块选择和静态回收块选择相结合的策略来选择目标回收块,根据回收块中有效页数据更新间隔判断有效页热度,同时提出数据更新稳定性的概念来划分有效页的数据更新状态,将具有不同热度和更新状态的有效页数据分别存储在不同的空闲块中,从而提高块中数据的同步更新概率。实验结果表明,UIGC算法相较于CAT、FaGC等现有垃圾回收算法具有更优的垃圾回收效率和磨损均衡效果,并能有效延长闪存使用寿命。

基于birch聚类的可更新机器学习索引模型

为满足大数据时代下数据库系统高吞吐、低内存占用的索引设计需求,提出一种面向海量数据的基于birch聚类可更新机器学习索引模型。将数据集使用birch聚类进行划分,对分段数据分别使用前馈神经网络进行训练拟合,采用基于日志结构合并树延迟更新思路的异地插入策略,实现索引更新操作。使用真实数据集进行实验,其结果表明,相比传统索引和当前先进机器学习索引结构,该模型在检索速度上有一定提升,在插入性能、内存占用和训练时间上有较大优化。

Long-term variabilities of thermodynamic structure of the East China Sea Cold Eddy in summer

Based on more than 30 years observed sectional temperature data since the 1960s, and compared with multi-year wind and Changjiang (Yangtze) River discharge data, spatial-temporal variations of the East China Sea Cold Eddy (ECSCE) in summer was analyzed in relationship to ocean circulation and local atmospheric circulation. Empirical Orthogonal Function (EOF) and Singular Value Decomposition (SVD) analyses were applied to this study. The results show that: 1) The ECSCE in summer possesses significant interannual variabilities, which are directly associated with oceanic and atmospheric circulation anomaly. Main fluctuations demonstrate their falling in basically with El Nino events (interannual) and interdecadal variability. 2) The ECSCE in summer is closely related to the variation of the Yellow Sea Warm Current (YSWC) and the Changjiang River discharge. The stronger the YSWC, the more intensive the ECSCE with its center shifting westward, and vice versa. However, a negative correlation between the Changjiang River discharge and the ECSCE strength is shown. The ECSCE was strengthened after the abrupt global climate change affected by the interdecadal variation of the YSWC. 3) SVD analysis suggested a high correlation between the variation of the ECSCE in summer and the anomalous cyclonic atmospheric circulation over the ECS. Intensification of the cyclonic wind strengthens the ECSCE, and vice versa. 4) The cyclonic atmospheric circulation has dominant influence on the interannual variation of the ECSCE, and the influence of the ocean circulation takes the second in. The ECSCE was usually stronger in El Nino years affected by strong cyclonic circulation in the atmosphere. The variation in strength of the ECSCE resulted from the joint effect of both oceanic and atmospheric circulation.

面向智慧城市的多云协同存储设计及实现

针对传统的单一云存储平台难以满足智慧城市领域中数据激增及其对数据高可靠性的要求,本文提出了多云协同存储系统方案,给出了实现海量数据高可靠的动态数据副本放置方法,旨在整合利用现有云存储平台实现面向智慧城市领域的大规模数据的协同存储,具有可扩展性、高效和低开销等优点。

Relationship between oceanic heat content and sea surface height on interannual time scale

The relationship between heat content and the interannual time scale is examined with satellite sea surface height （SSH） in the global ocean on altimeter measurements, historical hydrography, and model assimilation outputs. Results show that correlation between altimetric SSH and heat content in the upper 700 m calculated from Ishii data is geographically nonuniform. In the tropical ocean, heat content and SSH are strongly correlated and exhibit nearly the same interannual variations. In the polar ocean, their correlation is relatively weak. Further analysis with Simple Ocean Data Assimilation outputs shows that such nonuniform distribution is not from dynamical origin but from the limited integral depth selected to calculate heat content. The integral depth of 700 m is inadequate to capture variation of the deep main thermocline in the polar region. The halosteric effect also contributes to the nonuniform pattern of correlation, because saline contraction becomes significant in the polar ocean owing to low temperature.

Hydrography and Circulation in the Eastern Tropical Indian Ocean during April-May 2011

The distribution of hydrography and circulation in the eastern tropical Indian Ocean(ETIO) during April-May 2011 were analyzed using cruise observations,satellite observations,and historic hydrographic data.It was observed that warm water(>28℃) occupies the upper 50-m layer in the ETIO.Low-salinity surface water was observed at the mouth of the Bay of Bengal(BOB),which further extends to the Arabian Sea and off Sumatra via the Sri Lanka coast and the eastern bay mouth.Arabian Sea high-salinity water(ASHSW) is carried eastward along the equator to around 90°E by the equatorial undercurrent(EUC).It also runs south of Sri Lanka(north to 3°N) and in the western bay mouth(west to 87°E) but is much shallower than its counterpart at the equator.It is suggested to be the residual of the ASHSW,which intrudes into the BOB during the preceding southwest monsoon.Our results also show that,in the south of Sri Lanka,just below this subsurface high-salinity water,very-low-salinity water(about 34.8) occurs at depths of 100-200 m.Further analysis suggests that this low-salinity water comes from the BOB.

Satellite derived upper ocean thermal structure and its application to tropical cyclone intensity forecasting in the Indian Ocean

Upper ocean heat content is a factor critical to the intensity change of tropical cyclones(TCs). Because of the inhomogeneity of in situ observations in the North Indian Ocean,gridded temperature/salinity(T/S) profiles were derived from satellite data for 1993–2012 using a linear regression method. The satellite derived T/S dataset covered the region of 10°S–32°N,25°–100°E with daily temporal resolution,0.25°×0.25° spatial resolution,and 26 vertical layers from the sea surface to a depth of 1 000 m at standard layers. Independent Global Temperature Salinity Profile Project data were used to validate the satellite derived T/S fields. The analysis confirmed that the satellite derived temperature field represented the characteristics and vertical structure of the temperature field well. The results demonstrated that the vertically averaged root mean square error of the temperature was 0.83 in the upper 1 000 m and the corresponding correlation coefficient was 0.87,which accounted for 76% of the observed variance. After verification of the satellite derived T/S dataset,the TC heat potential(TCHP) was verified. The results show that the satellite derived values were coherent with observed TCHP data with a correlation coefficient of 0.86 and statistical significance at the 99% confidence level. The intensity change of TC Gonu during a period of rapid intensification was studied using satellite derived TCHP data. A delayed effect of the TCHP was found in relation to the intensity change of Gonu,suggesting a lag feature in the response of the inner core of the TC to the ocean.

Pool Boiling Heat Transfer in Dilute Water/Triethyleneglycol Solutions

Boiling of water/triethyleneglycol(TEG)binary solution has a wide-ranging application in the gas processing engineering.Design,operation and optimization of the involved boilers require accurate prediction of boiling heat transfer coefficient between surface and solution.In this investigation,nucleate pool boiling heat transfer coefficient has been experimentally measured on a horizontal rod heater in water/TEG binary solutions in a wide range of concentrations and heat fluxes under ambient condition.The present experimental data are correlated using major existing correlations.In addition a correlation is presented for prediction of pool boiling heat transfer for the system in which the vapour pressure of one component is negligible.This model is based on the mass transfer rate equation for prediction of the concentration at the bubble vapor/liquid interface.Based on this prediction,the temperature of the interface and accordingly,the boiling heat transfer coefficient could be straightforwardly calculated from the known concentration at the interface.It is shown that this simple model has sufficient accuracy and is acceptable below the medium concentrations of TEG when the vapor equilibrium concentration of TEG is almost zero.The presented model excludes any tuning parameter and requires very few physical properties to apply.

Long-term temperature variation of the Southern Yellow Sea Cold Water Mass from 1976 to 2006

This paper discusses the long-term temperature variation of the Southern Yellow Sea Cold Water Mass(SYSCWM)and examines those factors that infl uence the SYSCWM,based on hydrographic datasets of the China National Standard Section and the Korea Oceanographic Data Center.Surface air temperature,meridional wind speed,and sea surface temperature data are used to describe the seasonal changes.Mean temperature of the two centers of the SYSCWM had diff erent long-term trends.The temperature of the center in the west of the SYSCWM was rising whereas that of the center in the east was falling.Mean temperature of the western center was related to warm water intrusion of the Yellow Sea Warm Current,the winter meridional wind,and the winter air temperature.Summer process played a primary role in the cooling trend of temperature in the eastern center.A decreasing trend of salinity in the eastern half of the SYSCWM showed that warm water intrusion from the south might weaken,as could the SYSCWM circulation.Weakened circulation provided less horizontal heat input to the eastern half of the SYSCWM.Less lateral heat input may have led to the decreasing trend in temperature of the eastern center of the SYSCWM.Further,warmer sea surface temperatures and less heat input in the deep layers intensifi ed the thermocline of the eastern SYSCWM.A stronger thermocline had less heat fl ux input from upper layers to this half of the SYSCWM.Stronger thermocline and weakened heat input can be seen as two main causes of the cooling temperature trend of the eastern center of the SYSCWM.

Physicochemical properties of aqueous solutions of sodium glycinate in the non-precipitation regime from 298.15 to 343.15 K

The physicochemical properties,including the density,viscosity,and refractive index of aqueous solutions of sodium glycinate as a solvent for CO_2 absorption in the non-precipitation regime were measured under the wide temperature range of 298.15 to 343.15 K.The concentration of the sodium glycinate in an aqueous form in the non-precipitation regime was identified up to 2.0 mol ? L^(-1).The coefficients of thermal expansion values were estimated from measured density data.It was found that,the densities,viscosities and refractive indices of the aqueous sodium glycinate decrease with an increase in temperature,whereas with increasing sodium glycinate concentration in the solution,all three properties increase.Thermal expansion coefficients slightly increase with rising temperature and concentration.The measured values of density,viscosity and refractive index were correlated as a function of temperature by using the least squares method.The predicted data obtained from correlation equations for all measured properties were in fairly good agreement with the experimental data.

异构Hadoop集群中数据副本放置策略优化

针对默认的Hadoop数据副本策略未考虑集群节点硬件配置的异构、文件访问特点、实时负载等信息,导致异构环境中集群计算任务本地化比例下降、影响计算性能,提出计算型数据的副本放置优化策略.量化数据访问特征以及节点实时性能和负载,以节点数据访问负载与其计算性能相匹配为原则为副本选择存储节点.实验结果表明:与默认策略相比,优化的副本放置策略能更有效地为副本选择合适的存储节点,提高计算任务本地化比例和计算性能,并使集群对节点的变动具有更好的适应性.

磁光电混合存储技术研究综述

介绍了磁光电混合存储的技术特点及发展历程;总结了其关键技术及研究现状,包括存储系统结构、软件和硬件的关键技术以及相关标准的制定和专利申请;介绍了混合存储技术的发展动向,对磁光电混合存储的技术发展进行了展望。该综述有助于研究人员更系统、清晰、准确地认识磁光电混合存储技术,并有助于未来大数据存储的发展。

Thermospheric mass density derived from CHAMP satellite precise orbit determination data based on energy balance method

In this article, the energy balance method is used to retrieve thermospheric mass density from CHAMP satellite precise orbit determination(POD) data during 2007–2009. The retrieved thermospheric mass densities are compared with those from accelerometer data and an empirical model. The main conclusions are as follows:(1) Thermospheric mass density can be retrieved from POD data by the energy balance and semi-major axis decay methods, whose results are consistent.(2) The accuracy of the retrieved densities depends on the integration time period, and the optimal period for CHAMP density retrieval from POD data is about 20 minutes.(3) The energy balance method can be used to calibrate accelerometer data.(4) The accuracy of retrieving thermospheric density from POD data varies with satellite altitude and local time.