面向复杂中文地址关联的三维关系评估模型*

地址关系多样复杂阻碍信息业务发展,严重影响基于中文地址的大数据分析结果。针对此问题,调研地址关联工作并设计了一种三维关系评估模型。该模型根据地址关键元的文本属性、地理属性和统计属性分别计算其互相之间的直接关系、潜在关系和佐证关系取值,利用三维关系取值结果评估每一关键元“置信空间”中“相对率”最小的关系对做为关联关系。基于该模型实现了地址全部要素关联。实验证明此模型可有效提高地址关联率和准确率。     

Modeling the relationship between occupational stressors, psychosocial/physical symptoms and injuries in the construction industry

Injury statistics place the construction industry as a high-risk industry, making it necessary to investigate factors that influence accidents to be able to protect workers. Research was carried out to investigate the relationship existing among occupational stressors, psychological/physical symptoms and accident/injury and work days lost outcomes as experienced by manual workers engaged in a range of industrial construction occupations. Some of the occupational stressors significantly associated with self-reported and OSHA logged injuries were training, job certainty and safety climate of the company. The OSHA logged injuries were associated with the occurrence of headaches and feelings of tenseness on the job. These results imply that non-physical stressors should be included as a potential input associated with injuries in injury risk models for construction workers.

Relevance to industry

Traditional approaches to workers’ safety in the construction industry have focused on the physical and biomechanical aspects of work by improving tools, equipment and task completion methods. The impact of psychosocial factors, specifically stress as experienced by construction workers, is an area of growing research, which is yielding results that suggest overall work safety on the construction site should take into account psychosocial aspects of work.     

高效中药关联规则发现算法研究及应用

将关联规则发现算法引入到中药配方数据库的数据挖掘中,以求发现方剂中单方之间的关联规则及中药中的药对药组,可以为中药中新药的研制提供重要依据。由于常用的关联规则发现算法：Apriori算法存在多次扫描数据库的缺陷,提出了一种基于矩阵的关联规则发现算法：Apriori_Matrix算法,该算法优化了Apriori算法中集合连接过程多次比较所花费的时间,可极大地提高关联规则挖掘的效率。针对中药数据库中单方的种类有限、配伍规则各不相同、同一种病症对应多种方剂的情况,改进算法有助于缩短新药研制的周期。     

中药化学数据库关联规则的挖掘

数据挖掘是一种新兴的信息处理技术,本文将其中的关联规则运用到中药化学数据的处理,对其中的中医药效、植物科属、化学成分的活性、中药提取物现代药理等数据进行了维间关联规则的挖掘,找到了一系列的强规则,并对这些规则进行了分析,得到了其中有趣的关联规则,同时该关联规则的结果也说明了中药和西药在药效概念上的差异。该结果对于中药现代化,植物化学等相关的研究提供了一种新的思路。     

关联规则兴趣度研究

关联规则是一种常见的知识表达形式。本文介绍了关联规则的提取模式和基于PS架构提取模式的不足;介绍了关联规则兴趣度的定义,包括客观兴趣度和主观兴趣度以及综合兴趣度。     

基于归一化背景方向特征的脱机手写汉字识别

对4方向背景方向特征进行了改进,提出了8方向背景特征描述方法。与4方向背景方向特征描述方法相比,改进后的特征描述方法可以从0°、45°、90°、135°、180°、225°、270°、315°共8个方向来对汉字图像进行考察,从而进一步提高描述的精度。此外,为了消除笔划粗细的影响,还对背景方向特征进行了归一化处理。实验结果表明改进后的归一化8方向背景方向特征具有更高的识别精度。     

Compensation for occupational injuries and diseases: its effect upon prevention at the workplace

Abstract

In Switzerland, as in many other industrialized countries, the nature and extent of prevention at the workplace is determined, at least partially, by known cases of compensated occupational injuries and diseases. At both the national and international levels ∥ILO conventions) injuries and diseases that fit appropriate lists and definitions are eligible for compensation. It has been found, based upon an investigation of a representative sample (965 subjects) of the working population in the French-speaking region of Switzerland, that this restrictive view does not take into account the fact that a large proportion of injuries and diseases are claimed by the victims to be caused by their job. These injuries and diseases, responsible for at least one month's absence from work, are not considered to be eligible for compensation but must be covered by the patient's own insurance. Moreover, the survey showed that workers considered the ill effects on health and safety to be a consequence less of the physical working environment than of the work organization, and that this category of risks was not recognized. Thus, in addition to the reduction of hazards by the application of industrial hygiene, an informed improvement of the workplace and the work organization was required. Consequently, laws and regulations on occupational injuries and diseases should be changed in order to emphasize the role of more appropriate preventive tools, which includes ergonomics.     

一种有效的动态背景提取及更新方法

为了从复杂环境中获取高质量背景图像,提出一种基于帧间差分和背景减除法相结合的背景提取算法,采用多层次自适应背景更新的策略,能够克服新增物体对背景的影响,并能及时消除由于背景物体移出而造成的鬼影。实验证明该算法对光线变化鲁棒性高、速度快、更新效果好。     

Effects of text/background color combination,ambient illuminance,and display type on discriminating performance for young and elderly users

Abstract— This study examined the effects of text/background color combinations (black‐on‐white, red‐on‐white, green‐on‐white, and blue‐on‐white), ambient illuminance (50, 500, and 12,000 lx), and display type (color electronic‐paper displays and conventional transmissive liquid‐crystal displays) on the discriminating performance of young and elderly users. The results showed that when participants used the color e‐paper display, young participants had significantly better discriminating performance than elderly participants under the text/background color combinations of black‐on‐white, green‐on‐white, and blue‐on‐white, whereas no significant difference was present between young and elderly participants under red‐on‐white. When participants used a conventional transmissive LCD, however, there was no significant difference between young and elderly participants under all text/background color‐combination settings. Young participants had substantially better discriminating performance under 500 and 12,000 lx than under 50 lx, whereas elderly participants had the best discriminating performance under 12,000 lx, followed by that under 500 lx, and the worst performance under 50 lx. Participants who used a conventional transmissive LCD had substantially better discriminating performance under 500 and 12,000 lx than under 50 lx. Participants who used the color e‐paper display had the best discriminating performance under 12,000 lx, followed by that under 500 lx, and the worst performance under 50 lx. The results can be the guild for the designers of color e‐paper displays.     

一种基于关系矩阵的关联规则快速挖掘算法

首先对关联规则挖掘问题进行了简单的回顾,然后应用关系理论思想,引入了项目可辨识向量及其“与”运算,设计了一种快速挖掘算法SLIG,将频繁项目集的产生过程转化为项目集的关系矩阵中向量运算过程。算法只需扫描一遍数据库,克服了Aprori及其相关算法产生大量候选集和需多次扫描数据库的缺点。实验证明,与Aprori算法相比,SLIG算法提高了挖掘效率。     

一种自动提取中国书画作品中印章图像的方法

在分析了中国书画印章图像特点的基础上,针对基于内容的书画作品图像检索领域中存在的"语义鸿沟"问题,提出了一种自动提取中国书画作品中印章图像的方法,并设计实现了该算法.通过对提取效果进行分析,证明该方法有较高提取率,这对书画作品中图像语义和特定图像鉴别的研究有重要意义.     

一种有效的道路背景提取与更新算法

针对复杂环境中道路背景图像的快速获取问题,提出了一种快速有效的道路背景提取和更新算法。应用改进的多帧平均算法提取背景,采用改进的Surendra算法对背景进行更新。实验结果表明,该算法能够减轻初始静止车辆对背景建立的影响,能及时消除由于初始帧中目标移动而造成的鬼影,对光线变化鲁棒性高、速度快、更新效果好。     

一种消除PSD背景光干扰的新方法

光电位置敏感探测器(PSD)已应用到非接触式高精度动态位移测量中。但在实际应用中经常受到背景光的干扰,直接影响到测量系统的精度和可靠性。根据PSD的工作原理,分析了杂光下的输出信号与杂光的非线性关系。提出采用比例积分微分(PID)控制激光二极管的驱动电源来消除PSD背景光补偿方法,并设计了硬件补偿电路,实验结果和实际应用均表明:该方法能有效地消除背景光的干扰,可大大增加PSD的实用性。     

Effects of color display, color name, color formation and color alignment on the screen layout usability of customized product color combinations

This study investigated the effects of screen layout on the usability of a large number of color combination options for customized product. Using an L₁₈ orthogonal array, 18 interfaces were designed around four factors: color display, color name, color formation and alignment. Participants completed a color combination search task and a questionnaire concerning their satisfaction with the interfaces.Based on the search time and user satisfaction, the display of color samples by modular sections (itemized color chips) was found to be the optimal method of color selection. This arrangement reduced the number of combinations of color samples displayed on the screen and allowed efficient navigation of a large palette of color combinations. Vertical alignment aided the product color discrimination and reduced the search time. Users viewed various product color combinations directly with or without the aid of color names.

Relevance to industry

Because of the considerable observed reductions in search time and the superior reported degrees of satisfaction associated with the tested configurations for customized product color combination, the findings of this study have important ramifications for consumers and manufacturers of products as diverse as sofas, curtains, clothing, athletic shoes, cars, mobile phones and the other customized products.     

The occupational illness of slum dwellers across industries: A case study in West Bengal

Better livelihood of the slum dwellers necessitates analyzing the health status of the primary industrial earners upon whom the needs of the whole family rely. Studies concerned with the slum dwellers lack in addressing the occupational health of their primary earners who opt for various risky industrial occupations. The present study makes an empirical attempt in understanding the nature and extent of their occupational illness across the industries. The industries in the present study are categorized into four types, viz., civil and mechanical, textile, consumable items, and chemical. A multi-stage random sampling technique is used to collect the primary data from 320 industrial workers living in the slums of the Indian state West Bengal through face to face interviews and Focus Group Discussions (FGDs). The determinants of the nature of occupational illness are analyzed with an ordered Logit model. The chronic illness like respiratory diseases are prevalent among the workers of jute textile industries, and musculoskeletal disorders are found mostly among the workers of small garment and consumable items industries. The prevalence of occupational injuries is common among the workers of the civil-mechanical and chemical industries. The difference in the Out-of-Pocket Health Expenditure (OOPE) between doctor's consultation and consumption of Over the Counter (OTC) drugs of the industrial workers might prevent the workers from having a doctor's consultation in the long run. The provision and implementation of proper health scheme based on the exposure to health risks of the primary earners of slum household in the industries are suggested.     

改进的PSOGM算法在动态关联规则挖掘中的应用

针对动态关联规则挖掘中支持度向量和置信度向量变化趋势的分析和预测,提出一种改进的粒子群优化的灰色模型应用在动态关联规则挖掘中。由于灰色模型在引入背景值后导致在非平稳序列中的预测精度下降,因此有必要引入参数进行修正,通过在粒子群优化算法中引入二次搜索机制,优化求解灰色模型不同时刻的背景值,从而提高粒子群算法的局部搜索能力,进而提高灰色模型的预测精度。通过在Matlab平台上进行实验仿真,数据集采用超市购物数据,结果表明该方法比原始灰色模型、遗传算法优化的灰色模型和标准的粒子群优化的灰色模型具有更高的预测精度。     

一种改进的混合高斯模型背景估计方法

传统混合高斯模型一般为每个像素分配固定的高斯分布个数,从而造成背景形成速度的减慢和系统资源的浪费;同时也存在着高斯模型背景建模中的缓慢或滞留运动物体造成目标误判现象的问题(即空洞问题)。为此,提出了一种有效的两阶段视频图像处理方法。该方法在第一阶段根据像素点的优先级大小自动地调节高斯分布的数目,在第二阶段首先对像素点进行所属区域的划分,进而对目标区域和非目标区域采取不同的更新手段。实验表明,采用两阶段视频图像处理方法明显地改善了背景建模的速度,有效解决了提取目标出现的空洞问题。     

一种消除中文匹配中交集型歧义的方法

讨论一种消除中文匹配过程中所产生的交集型歧义的方法。该方法通过词典切分方法对中文文本进行切分,找出文本中的词语,作为理解中文的前提。对于匹配结果中出现的交集型歧义,通过计算匹配词语与其交集词语各自字符之间的互信息值来消除,互信息值较大的词说明该字符组合的可能性比较大。最后给出一个实际的消除歧义的例子,说明该方法的实用性。     

改进的道路背景提取和更新算法

现有算法对长时间含有静止车辆的视频序列进行背景提取和更新时效果不甚理想,为此提出一种改进的道路背景提取和更新算法。通过统计一段时间内视频帧中各像素值所属区间来提取背景图像,结合边缘检测算法分析持续变化的区域,解决背景更新中存在的当前背景图像中含有静止车辆的问题。实验表明算法可以很好地适应实际应用中的复杂情况,准确性有了明显提高,且算法实现简单,实时性较高。     

Effects of color sample display and color sample grouping on screen layout usability for customized product color selection

This study investigates the effects of color sample display and color sample grouping on the usability (task efficiency and user preference) of a color combination interface. A 180-subject nested design experiment tested each of the three levels of color sample grouping (associative color number, color series, and product section) against each of the two levels of color sample display (color chip, product thumbnail) for efficiency and effectiveness (search time and error rate). A separate 30-subject experiment evaluated user preference.