Data collection mode effect on feeling thermometer questions: A comparison of face-to-face and Web surveys

Feeling thermometer questions are widely used in political science research to estimate people’s attitudes and feelings toward a political object, like a political figure or an organization. Given the popularity of the feeling thermometer question in population surveys, more work is needed to explore the measurement of this question type. This study examines the data collection mode effect on feeling thermometers. Using the 2012 American National Election Studies, we find that the measurement of feeling thermometers is not exactly comparable between face-to-face and Web surveys. Face-to-face respondents tend to provide warmer feelings, while Web respondents give relatively more reliable responses in comparison. In both survey modes, respondents are most likely to select the response options that are verbally labeled although the effect is more striking in face-to-face than Web survey. The item nonresponse between these two modes does not differ in a meaningful way. This study ends by discussing future research directions on feeling thermometer questions.     

Distant Supervised Relation Extraction with Cost-Sensitive Loss

Recently, many researchers have concentrated on distant supervision relation extraction (DSRE). DSRE has solved the problem of the lack of data for supervised learning, however, the data automatically labeled by DSRE has a serious problem, which is class imbalance. The data from the majority class obviously dominates the dataset, in this case, most neural network classifiers will have a strong bias towards the majority class, so they cannot correctly classify the minority class. Studies have shown that the degree of separability between classes greatly determines the performance of imbalanced data. Therefore, in this paper we propose a novel model, which combines class-to-class separability and cost-sensitive learning to adjust the maximum reachable cost of misclassification, thus improving the performance of imbalanced data sets under distant supervision. Experiments have shown that our method is more effective for DSRE than baseline methods.     

基于动态标签的关系抽取方法

针对远程监督数据集的关系抽取研究方法存在着大量标签噪声的问题,提出了一种作用于分层注意力机制关系抽取模型的动态标签方法。首先，提出了一种根据关系类别相似性生成动态标签的概念。由于相同的关系标签包含相似的特征信息，计算特征信息的关系类别相似性有助于生成与特征信息相对应的动态标签。其次，利用动态标签方法的评分函数来评价远程监督标签是否为噪声，以决定是否需要生成新的标签代替远程监督标签，通过调整远程监督标签来抑制标签噪声对模型的影响。最后，根据动态标签来更新分层注意力机制以关注有效实例，重新学习每个有效实例的重要性，进一步抽取关键的关系特征信息。实验结果表明，相较于原始的分层注意力机制关系抽取模型，所提方法在Micro和Macro分数上分别有1.3个百分点和1.9个百分点的提升，实现了噪声标签的动态纠正，提升了模型的关系抽取能力。     

Towards patient‐centered packaging design: An industry perspective on processes,functions, and constraints

Medication packaging is essential to provide patients with guidance and correct use of their medicines for effective treatment. This research aims to increase knowledge about the medication packaging innovation process and its uptake towards patient‐centered packaging design. The study applied a qualitative research approach based on data from 25 in‐depth interviews with stakeholders involved in medication packaging design. The empirical data analysis revealed four themes that can improve and advance user‐centered packaging design: medication packaging innovation process, medication packaging functions and features, medication packaging design constraints, and patient‐centered medication packaging design. The findings suggest that medication packaging design is strongly affected by an emphasis on protective and safety packaging functions rather than on patients' needs. Packaging innovation usually is constrained by rigid incremental development processes, where compliance with regulations, extensive documentation, avoidance of manufacturing complexity, and considerations on cost prevail. These findings are discussed in relation to the three most evident trade‐offs for patient‐centered design: protection versus openability, utility versus cost, and complexity of manufacturability versus complexity of use. This research contributes with valuable input and additional evidence about the necessary shift to a user‐centered approach in a field that has not been design driven. This input complements previous research and provides an opportunity for industry decision makers and policy makers to lead patient‐centered packaging design that can benefit patients and relieve overloaded health care systems.     

基于GRU和注意力机制的远程监督关系抽取

随着深度学习的发展,越来越多的深度学习模型被运用到了关系提取的任务中,但是传统的深度学习模型无法解决长距离依赖问题;同时,远程监督将会不可避免地产生错误标签。针对以上两个问题,提出一种基于GRU（gated recurrent unit）和注意力机制的远程监督关系抽取方法,首先通过使用GRU神经网络来提取文本特征,解决长距离依赖问题;接着在实体对上构建句子级的注意力机制,减小噪声句子的权重;最后在真实的数据集上,通过计算准确率、召回率并绘出PR曲线证明该方法与现有的一些方法相比,取得了比较显著的进步。     

A Cost-Efficient Environment Monitoring Robotic Vehicle for Smart Industries

Environmental monitoring is essential for accessing and avoiding the undesirable situations in industries along with ensuring the safety of workers. Moreover, inspecting and monitoring of environmental parameters by humans lead to various health concerns, which in turn brings to the requirement of monitoring the environment by robotics. In this paper, we have designed and implemented a cost-efficient robotic vehicle for the computation of various environmental parameters such as temperature, radiation, smoke, and pressure with the help of sensors. Furthermore, the robotic vehicle is designed in such a way that it can be dually controlled by using the remote control along with the distant computer. In addition, contrary to the existing researches, the GSM modules are used to achieve the two-way long distance communication between the robotic vehicle and the distant computer. On the distant computer, the above-mentioned environmental parameters can be monitored along with controlling the robotic vehicle with the help of Graphical User Interface (GUI). In order to fulfill the given tasks, we have proposed two algorithms implemented at the robotic vehicle and the distant computer respectively in this paper. The final results validate the proposed algorithms where the above-mentioned environmental parameters can be monitored along with the smooth-running operation of the robotic vehicle.     

MLLR和MAP在远场噪声混响下的语音识别研究

自适应技术可以用较少的数据来调整声学模型参数,从而达到较好的语音识别效果,它们大多用于自适应有口音的语音。将最大似然线性回归（Maximum Likelihood Linear Regression,MLLR）、最大后验概率（Maximum A Posteriori,MAP）自适应技术用在远场噪声混响环境下来分析其在此环境下的识别性能。实验结果表明,仿真条件下,在墙壁反射系数为0.6,各种噪声环境下MAP有最好的自适应性能,在信噪比（Signal-to-Noise Ratio,SNR）分别为5 dB、10 dB、15 dB时,MAP使远场连续语音词错率（Word Error Rate,WER）平均降低了1.51%、12.82%、2.95%。真实条件下,MAP使WER下降幅度最大达到了37.13%。进一步验证了MAP良好的渐进性,且当自适应句数为1 000时,用MAP声学模型自适应方法得到的远场噪声混响连续语音的识别词错率比自适应前平均降低了12.5%。     

“异地多活”分布式存储系统设计和实现

随着互联网进入了大数据和云计算时代,分布式存储技术近年来受到了工业界和学术界的广泛关注。为了解决服务器压力过大、异地快速容灾和用户就近访问等问题,本文从异地多活、分布式CAP理论着手,分析了分布式存储的特性、应用场景和技术挑战,进而利用开源的Redis NoSql数据库,RabbitMQ消息队列等技术,搭建了一个满足最终一致性,可用性和分区容忍性的“异地多活”分布式存储系统。通过对系统的压力测试结果表明,基本可以满足工业级应用的吞吐量和一致性的要求。     

不同管径长距离输水干管的设计计算

对有两条平行敷设的长距离输水干管的流量分配、连通管间距的计算进行了研究，并提出了简化计算公式，指出此计算方法方便准确，可供长距离输水干管设计者参考。     

安徽国际会展中心空调通风设计

安徽国际会展中心总建筑面积5.2×104m2,是集展览、会议、餐饮、地下车库为一体的综合性建筑。文章从空调通风方式、气流组织、冷热源、空调自控等方面介绍了该工程空调通风设计。