游客情感计算的文本大数据挖掘方法比较研究

旅游文本大数据以其方便、快捷和低门槛的特点为游客情感计算提供了极大便利，已经成为旅游大数据的主要来源之一。基于大数据理论和情感理论，以文本大数据为数据源，在全面梳理国内外情感计算相关成果的基础上，利用人工智能中的逻辑/算法编程方法、机器学习方法、深度学习方法对旅游文本大数据进行挖掘，探索最佳的基于文本大数据的游客情感计算方法。研究发现：（1）基于情感词典的游客情感计算模型，其核心是构建情感词典和设计情感计算规则，方法简单，容易实现，适用语料范围广。（2）机器学习，用统计学方法抽取文本中的特征项，具有非线性特征，可靠性较线性特征的情感词典方法高。（3）基于深度学习技术的游客情感计算，效果良好，准确率在85%以上。训练多领域的文本语料易于移植，实用性强，且泛化能力好，较适合大数据时代游客情感计算研究。     

Evaluation of green and efficient deep eutectic solvents as media for extracting alkaloids from lotus leaf

Deep eutectic solvents (DESs) were applied as eco-friendly solvents in this study for the extraction of alkaloids from lotus leaf, including O-nornuciferine, N-nornuciferine, nuciferine and roemerine. A series of hydrophilic and hydrophobic DESs with different hydrogen bond donors and a acceptors were synthesized and screened for a suitable DESs for extraction of alkaloids from lotus leaf. The study results showed that the hydrophilic DES with choline chloride and propanediol had the highest extraction yield. The main factors affecting the extraction efficiency—choline chloride–propanediol ratio, water content in deep eutectic solvents, solid–liquid ratio and extraction time—were investigated via a single-factor experiment. The optimized extraction conditions were 30% of water in choline chloride–propanediol (1:4) for heated extraction for 30 min and solid–liquid ratio 1:100 g/ml. Under optimum conditions, the extraction yields of O-nornuciferine, N-nornuciferine, nuciferine and roemerine were 0.069, 0.152, 0.334 and 0.041 g/100 g respectively, which were higher than those of methanol in acidified aqueous solution. This study suggests considerable potential for DESs as promising materials for the green and efficient extraction solvents for bioactive alkaloids from natural sources.     

RNAGCN: RNA tertiary structure assessment with a graph convolutional network

RNAs play crucial and versatile roles in cellular biochemical reactions. Since experimental approaches of determining their three-dimensional (3D) structures are costly and less efficient, it is greatly advantageous to develop computational methods to predict RNA 3D structures. For these methods, designing a model or scoring function for structure quality assessment is an essential step but this step poses challenges. In this study, we designed and trained a deep learning model to tackle this problem. The model was based on a graph convolutional network (GCN) and named RNAGCN. The model provided a natural way of representing RNA structures, avoided complex algorithms to preserve atomic rotational equivalence, and was capable of extracting features automatically out of structural patterns. Testing results on two datasets convincingly demonstrated that RNAGCN performs similarly to or better than four leading scoring functions. Our approach provides an alternative way of RNA tertiary structure assessment and may facilitate RNA structure predictions. RNAGCN can be downloaded from https://gitee.com/dcw-RNAGCN/rnagcn.     

虚拟震源测深法及其在INDEPTH-3台阵中的应用

基于中美合作项目INDEPTH第3期在青藏高原布设的台站，使用虚拟震源测深法研究青藏高原中部的地壳厚度。结果显示，拉萨地体和羌塘地体的地壳结构存在巨大差异。拉萨地体的地壳厚度大约为57 km，与艾里均衡说预测的地壳厚度基本一致，说明拉萨地体的地壳结构比较简单。羌塘地体的地壳厚度为60~75 km，向北有增厚趋势，明显较艾里均衡说预测的地壳厚，说明羌塘地体地壳结构比较复杂，原因有可能是羌塘地体下存在高温流体和低速带，或者与印度板块岩石圈在班公湖-怒江缝合带以北向下俯冲有关。     

Enhanced headspace single drop microextraction method using deep eutectic solvent based magnetic bucky gels: Application to the determination of volatile aromatic hydrocarbons in water and urine samples

A facile headspace single drop microextraction method was developed using deep eutectic solvent‐based magnetic bucky gel as the extraction solvent for the first time. The hydrophobic magnetic bucky gel was formed by combining choline chloride/chlorophenol deep eutectic solvent and magnetic multiwalled carbon nanotube nanocomposite. Magnetic susceptibility, high viscosity, high sorbing ability, and tunable extractability of organic analytes are the desirable advantages of the prepared gel. Using a rod magnet as a suspensor in combination with the magnetic susceptibility of the prepared gel resulted in a highly stable droplet. This stable droplet eliminated the possibility of drop dislodgement. The prepared droplet made it possible to complete the extraction process in high temperatures and elevated agitation rates. Furthermore, using larger micro‐droplet volumes without any operational problems became possible. These facts resulted in shorter sample preparation time, higher sensitivity of the method, and lower detection limits. Under the optimized conditions, an enrichment factor of 520–587, limit of detection of 0.05–0.90 ng/mL, and linearity range of 0.2–2000 ng/mL (coefficient of determination = 0.9982–0.9995) were obtained. Relative standard deviations were < 10%. This method was successfully coupled with gas chromatography and used for the determination of benzene, toluene, ethylbenzene, and xylene isomers as harmful volatile organic compounds in water and urine samples.     

Semiotic Aggregation in Deep Learning

Convolutional neural networks utilize a hierarchy of neural network layers. The statistical aspects of information concentration in successive layers can bring an insight into the feature abstraction process. We analyze the saliency maps of these layers from the perspective of semiotics, also known as the study of signs and sign-using behavior. In computational semiotics, this aggregation operation (known as superization) is accompanied by a decrease of spatial entropy: signs are aggregated into supersign. Using spatial entropy, we compute the information content of the saliency maps and study the superization processes which take place between successive layers of the network. In our experiments, we visualize the superization process and show how the obtained knowledge can be used to explain the neural decision model. In addition, we attempt to optimize the architecture of the neural model employing a semiotic greedy technique. To the extent of our knowledge, this is the first application of computational semiotics in the analysis and interpretation of deep neural networks.     

基于多尺度学习与深度卷积神经网络的遥感图像土地利用分类

土地利用信息是国土资源管理的基础和重要依据，随着高分辨率遥感图像数据的日益增多，迫切需要快速准确的土地利用分类方法。目前应用较广的面向对象的分类方法对空间特征的利用尚不够充分，在特征选择上存在一定的局限性。为此，提出一种基于多尺度学习与深度卷积神经网络（deep convolutional neural network，DCNN）的多尺度神经网络（multi-scale neural network，MSNet）模型，基于残差网络构建了100层编码网络，通过并行输入实现输入图像的多尺度学习，利用膨胀卷积实现特征图像的多尺度学习，设计了一种端到端的分类网络。以浙江省0.5 m分辨率的光学航空遥感图像为数据源进行了实验，总体分类精度达91.97%，并将其与传统全卷积网络（fully convolutional networks，FCN）方法和基于支持向量机（support vector machine，SVM）的面向对象方法进行了对比，结果表明，本文所提方法分类精度更高，分类结果整体性更强。     

基于分子工程的方法设计首例具有Sr2Be2B2O7 (SBBO)结构的深紫外氟碳酸盐双折射晶体AMgLi2(CO3)2F (A=K,Rb)※

分子结构设计是开发新化合物和通过原子尺度操纵优化晶体结构的一种引人注目的策略. 在这个工作中, 利用分子工程的思想, 基于SBBO结构, 成功设计并合成两个新型氟碳酸盐KMgLi₂(CO₃)₂F和RbMgLi₂(CO₃)₂F. 在两个结构中, a-b平面是由CO₃和LiO₃F阴离子基团组成的无限[Li₃C₃O₆F₃]_∞层, 进一步相邻的层通过F原子连接形成一个独特的[Li₆C₆O₁₂F₃]_∞双层. 这种结构特征对改善晶体的层状生长习性和消除晶体的多晶性有很大的帮助. 光学测试表明, 该系列晶体具有大的双折射和短的紫外截止边, 是深紫外双折射晶体良好的候选材料.     

医学影像处理的深度学习可解释性研究进展

随着医学影像数据的迅速增长，传统的影像分析方法给医生带来巨大挑战。利用计算机视觉技术提供自动或半自动辅助诊断，可大大缓解人工阅片压力，提高诊断的准确性，促进医疗流程的标准化建设等。目前，深度学习卷积神经网络在医学影像处理中已取得不俗表现，但深度学习“黑匣子”的不可解释性阻碍了智能医疗诊断的发展。为增强对医学影像数据处理的深度学习可解释性的了解，对近几年相关研究进展进行了综述。首先，综述了深度学习在医学领域的应用现状及面临的问题，对神经网络的可解释性内涵进行了讨论；然后，从现有深度学习可解释性的常见方法出发，重点讨论了医学影像处理的深度学习可解释性研究进展；最后，探讨了医学影像处理的深度学习可解释性的发展趋势。     

融合图卷积神经网络和注意力机制的PM2.5小时浓度多步预测

PM_2.5小时浓度多为单步预测。为实现PM_2.5小时浓度的多步预测，基于“编码器-解码器”的序列-序列预测（Seq2Seq）模型，集合图卷积神经网络提取非欧式空间数据特征的能力以及注意力机制自适应关注特征的能力，提出了融合图卷积神经网络和注意力机制的PM_2.5小时浓度多步预测（GCN_Attention_Seq2Seq）模型。并与Seq2Seq模型和使用了图卷积神经网络、未使用注意力机制的GCN_Seq2Seq模型进行了对照，以2015—2016年北京市22个空气质量监测站点的空气质量数据为样本进行实例验证，结果表明，Seq2Seq模型和图卷积神经网络（GCN）可对PM_2.5小时浓度数据的时空依赖进行有效建模，注意力机制有助于减缓多步预测中的预测精度衰减，提升PM_2.5小时浓度多步预测的精度。GCN_Attention_Seq2Seq模型可有效应用于多种长度的PM_2.5浓度预测窗口。