Feature selection based on mutual information and redundancy-synergy coefficient

INTRODUCTION Feature subset selection (FSS) is a data miningfundamental problem to select out relevant featuresand cast away irrelevant and redundant featuresfrom an original feature set (Liu and Motoda, 1998).If a feature subset satisfies the FSS measure and hasthe minimal size, it is regarded as the optimal fea-ture subset. Complete search strategy is the way toobtain an optimal feature subset. Branch and Bound(Narendra and Fukunaga, 1977), Focus (Almuallimand Dietterich, 199…     

国内外健康医学科学数据管理平台对比分析

基于科学数据管理生命周期理论与科学数据管理能力成熟度模型,从建设基础(政策、经费、技术、团队)及管理过程(数据管理计划、采集标准与流程、描述与元数据、存储与保存、共享与保护、应用与服务)两个维度对比国内外11个健康医学科学数据管理平台的建设实践。调查发现,国内健康医学科学数据管理政策不够完善,缺少保障健康医学数据安全及个人隐私的法规;平台提供的服务强调健康医学科学数据存储,而对数据管理与共享的支持不足。基于国外成熟经验,本文从完善激励机制与数据政策、扩大经费来源、增强技术创新、保护数据安全和促进数据共享、深化服务内容5个方面对国内健康医学科学数据管理平台提出建议。     

Data management maturity assessment of public sector agencies

To determine how critical data assets are conceptualized and managed in the public sector, we conducted a large-scale empirical study at 15 government agencies. We use the Data Management Maturity (DMM) reference model framework to conduct a systematic multi-level analysis (inter-agency, intra-agency, and cross-case analysis). To aid the comparative assessment of multiple independent agencies, we propose and test the DMM Index. The study not only examines the maturity of data management practices in government agencies, but also provides guidance on how an enterprise-wide, systematic assessment may be conducted. The approach presented in the paper can be replicated at other large government entities and private conglomerates. Public and private sector agencies may apply the approach to develop custom roadmaps for data management improvements that align with the organization's business goals.     

数字校园中的数据交换平台设计

数字化校园建设是一个复杂的系统工程,面对的一个现实问题就是需要在不同业务系统之间实现数据集成与交换。本文基于海军工程大学网络综合应用平台项目的建设经验,提出了数字校园数据交换平台的体系结构,并对应用适配器、数据交换引擎、Web服务器等关键技术进行了设计和实现。     

Kinetic analysis of γ-glutamyltransferase reaction process for measuring activity via an integration strategy at low concentrations of γ-glutamyl p-nitroaniline

At 0.12 mmol/L γ-glutamyl p-nitroaniline (GGPNA), an improved integrated method was developed for kinetic analysis of γ-glutamyltransferase (GGT) reaction process and the integration with the classical initial rate method to measure serum GGT. For the improved integrated method, an integrated rate equation, which used the predictor variable of reaction time and considered inhibitions by both GGPNA and products, was nonlinearly fit to GGT reaction processes. For the integration strategy, classical initial rates were estimated when GGPNA consumption percentages were below 50%; otherwise, maximal reaction rates of GGT were estimated by the improved integrated method and converted into initial rates according to the differential rate equation at 0.11 mmol/L GGPNA. The integration strategy was validated using optimized GGT kinetic parameters and 10-s intervals to record reaction curves within 8.0 min. By the integration strategy, there was a linear response from 0.9 to 32.0 U/L GGT, coefficients of variation were below 3.5% for GGT from 8.0 to 32.0 U/L (n=5), and GGT activities in clinical sera responded linearly to their classical initial rates at 2.00 mmol/L GGPNA with an expected slope. Therefore, the integration strategy was successful in measuring GGT at 0.12 mmol/L GGPNA.     

数据挖掘技术在高校教学质量评价中的应用

针对高校教学质量评价,利用数据挖掘技术,运用多种数据挖掘方法对教学相关数据进行分析,以期达到改进教学方法,提高教学质量的目的。     

AHP和DEA相结合的高校绩效综合评价模型研究

基于投入／产出的DEA方法是高校绩效／效率评价的常用经典方法,但该方法在指标数量和原始数据方面有一定的限制。为了解决上述问题,文章将AHP和DEA相结合构建高校绩效综合评价模型。首先,将定性数据用AHP进行量化。其次,应用AHP计算二级指标权重,加权平均计算一级指标值作为输入输出数据,减少指标数量,使之符合DEA要求的较少的指标数目。最后,基于DEA基本模型和超效率模型评价高校绩效并加以分析。     

动态时间弯曲距离精确计算的Dijkstra方法

将动态时间弯曲距离（DTW）的差异矩阵一一对应于点阵,按DTW定义的行走规则对该点阵连线定向,使所对应点阵成为一个有向图,然后使用一个加权技巧对该有向图的边加权后得到一个加权有向图,于是把求DTW的精确计算问题等价地转化为求一个有向图起点到终点的最短路长,从而使图论中求两点间最短路径的方法如目前公认的经典Dijkstra算法均可用于求DTW,因此间接地找到了精确计算DTW的一个新方法.     

The role of positive and negative valence factors on the impact of bigness of data on big data analytics usage

The number of firms that intend to invest in big data analytics has declined and many firms that invested in the use of these tools could not successfully deploy their project to production. In this study, we leverage the valence theory perspective to investigate the role of positive and negative valence factors on the impact of bigness of data on big data analytics usage within firms. The research model is validated empirically from 140 IT managers and data analysts using survey data. The results confirm the impact of bigness of data on both negative valence (i.e., data security concern and task complexity), and positive valence (i.e., data accessibility and data diagnosticity) factors. In addition, findings show that data security concern is not a critical factor in using big data analytics. The results also show that, interestingly, at different levels of data security concern, task complexity, data accessibility, and data diagnosticity, the impact of bigness of data on big data analytics use will be varied. For practitioners, the findings provide important guidelines to increase the extent of using big data analytics by considering both positive and negative valence factors.     

基于知识管理的数据服务企业商业模式分析

本文首先回顾了商业模式概念、知识管理理论以及数据服务行业的相关研究,并以数据服务行业中具有代表性的初创企业杭州F科技有限公司为例,从价值主张、业务系统、盈利模式、关键资源能力以及外部环境等五大方面对其显性知识和隐性知识进行分析,挖掘其成功原因。研究发现数据服务类企业创业初期成败的关键是在商业模式中平衡业务系统与盈利模式,即找准市场明确定位并且维护良好政企关系,通过自身关键资源能力优势参与行业标准制定,从而树立行业壁垒获得竞争优势。