Human resource management was of great significance for enterprises to formulate development strategies and improve competitiveness. With the rapid development of new technologies such as knowledge economy and big data, the traditional human resource performance appraisal model had certain limitations and was not objective, which had been difficult to adapt to the increasingly highly integrated development needs of modern enterprises. Therefore, this paper used big data analysis and new technical means to put forward the study of human resource performance evaluation model based on Bayesian network, aiming to provide some theoretical reference for modern enterprise human resource management. First of all, it summarized the related concepts of performance appraisal and its important position in enterprise management, analyzed the characteristics of human resources performance appraisal indicators and their mutual relations, and expounded the common methods and characteristics needed to build a performance appraisal system. Secondly, it summarized the relevant theory of Bayesian network and its advantages in solving complex random problems, and proposed a performance appraisal model based on Bayesian network for the needs of enterprise human resources management. Finally, the balanced scorecard method was used to build a performance appraisal system, and the performance appraisal model proposed in this paper was tested by experiments, and compared with the traditional method. The results showed that the performance appraisal model proposed in this paper had significant advantages and could be better applied to the performance appraisal management of enterprise human resources. The research of this paper can not only provide theoretical guidance for further research on enterprise human resource allocation and management, but also provide reference for performance appraisal in other fields.

脑电波采集过程往往会包含生理噪声和外部噪声,外部噪声目前可以通过滤波进行消除,去除生理噪声通常采用的方法是自适应滤波器、空间滤波和主成分分析,但这些方法需要一定的信号作为先决条件,并且去噪性能有限.近年来,深度学习技术开始运用于脑电波去噪,而且在去噪性能上得到了一定提升.鉴于在生成对抗网络(GAN)的判别器和生成器博弈过程中,生成样本具有逐渐向真实样本逼近的特点,设计一种基于卷积神经网络(CNN)的生成对抗网络模型,用于消除脑电波生理噪声中的肌源性噪声和眼源性噪声,同时在GAN的生成器网络中引入新的损失函数,使去噪后的数据与原始数据更加接近.设计的GAN-1D-CNN模型在去除肌源性噪声上的相关系数达到0.945,在去除眼源性噪声上的相关系数达到0.894.实验结果表明,GAN-1D-CNN模型对脑电波的去噪能力得到了增强,相关性能指标都优于现有基准数据集上的去噪方法.

针对目前大多数脑电波情感识别方法存在的依赖手动特征提取等问题,提出一种基于卷积神经网络(convolutional neural network,CNN)和双向长短时记忆(bidirectional long short-term memory,Bi-LSTM)网络的混合模型.首先将一维数据转换为二维数据,采用CNN提取空间特征;然后将一维数据输入Bi-LSTM,获取时间特征;最后将融合的空间和时间特征输入Softmax分类器,得到最终分类结果.在DEAP数据集上的实验结果表明:CNN和Bi-LSTM混合模型具有较好的分类性能,在效价度和唤醒度上的准确率分别达到88.55％和89.07％,是一种可行的脑电波情感分类模型.

情感分析作为自然语言处理(NLP)的细分研究方向经历了使用情感词典、机器学习和深度学习分析的发展过程.针对使用一般化的深度学习模型作为文本分类器对于特定领域的网络评论类型的文本的分析的精准度较低,训练时发生过拟合现象以及情感词典覆盖率低、编纂工作量大的问题,提出了基于情感词典和堆叠残差的双向长短期记忆(Bi-LSTM)网络的情感分析模型.首先,借助情感词典中情感词的设计覆盖"教育机器人"研究领域内的专业词汇,从而弥补Bi-LSTM模型在分析此类文本时精准度的不足;然后,使用Bi-LSTM和SnowNLP来降低情感词典的编纂体量.长短期记忆(LSTM)网络的"记忆门""遗忘门"结构可以在保证充分考虑评论文本中的前后词语的关联性的同时,适时选择遗忘一些已分析词语,从而避免反向传播时的梯度爆炸问题.而在将堆叠残差的Bi-LSTM引入后,不仅使得模型的层数加深至8层,而且还使残差网络避免了叠加LSTM时会导致的"退化"问题;最后,通过适当设置和调整两部分的得分权重,并将总分使用Sigmoid激活函数标准化到[0,1]的区间上,按照[0,0.5],(0.5,1]的区间划分分别表示负面和正面情绪,完成情感分类.实验结果表明,在"教育机器人"评论数据集中,所提模型对于情感分类准确率相较于标准的LSTM模型提升了约4.5个百分点,相较于BERT提升了约2.0个百分点.综上,所提模型将基于情感词典和深度学习模型的情感分类方法一般化;而通过修改情感词典中的情感词汇并适当调整深度学习模型的结构和层数,所提模型可以应用于电子商务平台中各类商品的购物评价的精确情感分析,从而帮助企业洞悉消费者的购物心理和市场需求,同时也可以为消费者提供商品质量的一种参考标准.

[目的/意义]旨在为促进科技知识传播提供参考.[方法/过程]构建了多层耦合网络的传播模型,根据目前全国科技馆的情况,研究了科技知识在多层网络中的传播情况,并对影响传播的多种因素进行了仿真实验.[结果/结论]网上虚拟科技馆在多层网络中能加速科技知识的传播;科技馆影响力影响科技信息传播速度;用户节点之间兴趣匹配关系直接影响知识传播的速度和范围.

At present, most deep learning-based analysis of student’s attentional states in class has been studied only for a single model structure, and there is not enough recognition accuracy. To address this issue, an attention classification model FF-BiALSTM is proposed, which integrates an Attention Mechanism and a bi-directional long short-term memory neural network (Bi-LSTM). The Attention Mechanism is used to capture global features better and two Bi-LSTM layers are employed to capture time-domain features more effectively. This study defined two attention states to identify whether students are focused or not. Experiments on the Student EEG and Student Reading datasets show that this algorithm can effectively improve student attention classification performance. This experiment obtained 97.77% accuracy on the Student EEG training set and 91.35% on the Student EEG testing set.

针对目前对话文本情感分析中大部分模型忽略说话者情感的相互影响作用这一问题,为了有效识别对话文本中说话者本身表达的情感类别,充分考虑对话者之间的情感因素,提出一种基于图卷积神经网络的对话情感分析方法.首先,使用BiGRU将对话文本进行序列上下文编码,获得话语文本表征;然后,依据说话者对话顺序构造一个有向图,利用图卷积神经网络获取每个话语文本新的文本表征向量;最后,连接得到的两个话语表征向量,采用基于相似度的注意力机制获得最终的话语文本表示,从而进行情感分类.在dailydialog中文语料库上的实验结果表明,BiGRU结合GCN模型的方法相比于CNN和BiLSTM模型,在对话文本情感分类方面的准确率大约提高了15％,且F1值也有明显提高,取得较好的情感分类效果.

Internet of Medical Things (IoMT) is a kind of Internet of Things (IoT) that includes patients and medical sensors. Patients can share real-time medical data collected in IoMT with medical professionals. This enables medical professionals to provide patients with efficient medical services. Due to the high efficiency of cloud computing, patients prefer to share gathering medical information using cloud servers. However, sharing medical data on the cloud server will cause security issues, because these data involve the privacy of patients. Although recently many researchers have designed data sharing schemes in medical domain for security purpose, most of them cannot guarantee the anonymity of patients and provide access control for shared health data, and further, they are not lightweight enough for IoMT. Due to these security and efficiency issues, a novel lightweight privacy-preserving data sharing scheme is constructed in this paper for IoMT. This scheme can achieve the anonymity of patients and access control of shared medical data. At the same time, it satisfies all described security features. In addition, this scheme can achieve lightweight computations by using elliptic curve cryptography (ECC), XOR operations, and hash function. Furthermore, performance evaluation demonstrates that the proposed scheme takes less computation cost through comparison with similar solutions. Therefore, it is fairly an attractive solution for efficient and secure data sharing in IoMT.

针对中国历史典故传统展示模式存在的交互性不足、内容枯燥等问题,设计并实现了基于AR技术的中国历史典故展示系统.采用3ds Max创建虚拟模型,采用OpenCVForUnity实现基于SIFT-BRISK算法的三维注册,利用Unity3D引擎搭建虚拟场景,实现了多维度、跨时空的中国历史典故交互展示.实验结果表明,系统运行平稳流畅、鲁棒性强,能够有效提升中国历史典故展示的趣味性与体验感.

多视点视频是指在场景中放置多台摄像机,记录下多个视点数据,提供给用户视点选择和场景漫游的交互式媒体应用.多个摄像机从不同视角同时拍摄同一场景得到的一组视频信号,每一个摄像机代表一个不同的视角.可同时传输多个空间角度的视频流到用户端,并合成用户所需要的视域图像.多视点视频是一种新型的具有立体感和交互操作功能的视频,是未来一种极具应用前景的多媒体应用.然而,当前多视点的无线网络带宽分配机制中,都没有考虑大量的、不同解码能力的用户共存时的效率问题.一般情况下,合成视域往往需要至少左右两边(两条以上)的参考视频同时传输到用户端,才能使合成的视域质量不低于直接传输的视点质量,使得网络数据量成倍增加.同时,用户的设备性能影响用户感知质量.网络中手机屏幕和高清大屏显示对网络传输视频数据率的要求不同,必须考虑用户设备解码能力的限制,才能真正提供用户满意的感知质量.本文通过考虑移动无线网络带宽约束,考察不同用户端硬件的解码能力、视域大小及带宽消耗,利用博弈理论,使整体网络资源收益最大.本文分别考虑了几种特定场景下多视点视频传输的无线网络的资源分配.第一,已知网络用户的满意度参数,不考虑带宽的限制(带宽充足),如何确定每个用户需要支付的单位价格.第二,在用户个数不确定的条件下,如何判断出可以接入的用户个数和用户需提供的单位价格.第三,同时考虑用户设备对最大场景复杂度解码能力受限以及网络带宽受限两个约束条件,同时进行用户接入控制和多视点的视频质量优化,使得网络的收益和用户的效用得到最大化.本文对提出的算法进行了理论分析,证明了本文参数设置的合理性.在多视点移动网络资源调度中,本文提出的算法可以方便设置所需的价格参数.从视域大小、价格、用户效用、网络收益等各方面对实验性能进行比较.仿真结果显示本方法在同等实验条件下,多视点用户效用提升分别为5％和12％,网络总体收益增加32％.本文算法可以同时满足网络收益和用户整体效用最优,提高多视点视频在多用户下的网络资源利用率.

传统的Apriori算法要多次扫描数据集,随着数据量的快速增长,传统的Apriori算法已经不能很好地适用于大数据分析,针对该情况设计了IPApriori算法.首先通过剪枝策略设计了一种适用于多维数据的IApriori算法,再将IApriori算法与Hadoop分布式框架相结合,实现了多维关联规则挖掘算法的并行化.将IPApriori算法运用到手机用户行为预测关联分析中,分析影响手机用户行为的一些主要因素,挖掘出手机用户行为与年龄维度、性别维度、时间维度、地点维度和手机品牌维度属性之间可能存在的某种关联.最后通过实验证明,算法的并行化和建立结构的方法可以降低系统的I/O负荷,提高算法的执行效率.

To improve the management of science and technology museums, this paper conducts an in-depth study on Wi-Fi (wireless fidelity) indoor positioning based on mobile terminals and applies this technology to the indoor positioning of a science and technology museum. The location fingerprint algorithm is used to study the offline acquisition and online positioning stages. The positioning flow of the location fingerprint algorithm is discussed, and the improvement of the location fingerprint algorithm is emphasized. The raw data of the RSSI (received signal strength indication) is preprocessed, which makes the location fingerprint data more effective and reliable, thus improving the positioning accuracy. Three different improvement strategies are proposed for the nearest neighbor classification algorithm: a balanced joint metric based on distance weighting and a compromise between the two. Then, in the experimental simulation, the positioning results and errors of the traditional KNN (k-nearest neighbor) algorithm and three improvement strategy algorithms are analyzed separately, and the effectiveness of the three improved strategy algorithms is verified by experiments.

With the development of the Internet industry and the smart phone industry, the amount of mobile application software is increasing and the variety is becoming more and more abundant. The ensuing question is how mobile users can choose the applications they are interested in among many mobile applications. Therefore, mobile application recommendation service for mobile phone users came into place. This paper takes mobile phone users as the starting point, uses multiple linear regression models, uses the commonly used least mean square to evaluate the error, and uses the batch gradient descent method to optimize the training analysis error. Through experiments on the usage data of a large number of mobile phone users, the relationship between gender, age, mobile phone model, mobile phone usage location and app category of the mobile phone user is analyzed, and the mobile phone user app category is predicted to personalized recommend apps for the mobile phone user.

This paper presents a personalized course recommended algorithm based on the hybrid recommendation. The recommendation algorithm uses the improved NewApriori algorithm to implements the association rule recommendation, and the user-based collaborative filtering algorithm is the main part of the algorithm. The hybrid algorithm adds the weight to the recommendation result of the user-based collaborative filtering and association rule recommendation, implementing a hybrid recommendation algorithm based on both of them. It has solved the problem of data sparsity and cold-start partially and provides a academic reference for the design of high performance elective system. The experiment uses the student scores data of a college as the test set and analyzes results and recommended quality of personalized elective course. According to the results of the experimental results, the quality of the improved hybrid recommendation algorithm is better.

陶渊明的隐居与沈约的郊居存在极大的相似性,引入沈之郊居作为参照点,有利于更加深入地观照陶之隐居.沈约接受了佛教思想,以郊居来排遣俗累,以佛国净土的美妙幻想来消解对死亡的恐惧.陶渊明间接地受到了佛教的影响,放弃了佛道解脱法,直面死亡的残酷,在隐居之中完成了对名利之心的消解,对新的人生价值的寻找和确认,以及对死亡恐惧的坚强博弈与最终超越.

A novel logistic binary matrix factorization (LBMF) was proposed to predict the students' performance and to classify the exam items.Besides a new algorithm was designed to tackle the non-convex optimization problem involved in LBMF.The experiments are performed on both simulated data and real data.The results indicate that LBMF can not only predict the students' academic performance but also classify the examination items according to the knowledge points they require.And it can be concluded that LBMF outperforms significantly the out-of-date algorithms in the applications.

The existing parallel FP-Growth algorithm have solved the problems such as the partition of transaction dataset, which can guarantee that each transaction dataset is independent after the partition, but there are still many problems such as too many iterations in the process of FP-tree mining on single node and low efficiency. What’s more, it did not consider the load balance when the master node divides dataset to the child nodes. By using Cutting strategy on PFP which is the original algorithm with MapReduce, we merged paths which are not frequent in FP-tree, and designed a new parallel FP-Growth algorithm. In addition, the load balancing strategy is used when the master nodes divide dataset to the child nodes. Through the combination of these two strategies, this paper designed a new parallel FP-Growth algorithm.

By adopting graph theory,congestion control of express network is studied in this paper.Through the analysis of the characteristics of the network traffic flow and the study on the effect of the structure of express network on the network transmission capability,balancing the relationship between the network transmission capability and the connection cost.First of all,the concept of betweenness is introduced.Considering the relationship between the betweenness and cargo flow,the betweenness definition is modified,and the calculation method of betweenness is designed.Next,according to the betweenness calculation formula,the relationship of express network transmission capacity,node betweenness and node capacity are derived.Then,by taking the minimum connection cost as the optimization goal,an optimization model of express delivery network with the constraint of expect transmission capacity is constructed,and an algorithm is designed to seek the network with the optimal structure by gradually adding edge,reconnecting edge and deleting edge.Finally,the example of the backbone network of an express delivery company in Guangxi province is taken to verify the effectiveness of the model and algorithm.The result of simulation indicates that the algorithm can effectively find out the optimal delivery network.Through the research,it is found that processing power and betweenness of the bottleneck node decision network transmission capacity,and there is a contradiction between network transmission capacity and connection cost.

This paper aims at making optimization on express delivery network based on complex network theory,which changes the previous optimization method based on hub-and-spoke network theory.First,it draws on delivery timeliness to characterize network delivery efficiency and the total length of network connection edges to characterize connection cost.Next,taking the minimum connection cost as the optimization goal,it constructs an optimization model of express delivery network with the constraint of delivery timeliness.Then,it designs an algorithm to seek the network with the minimum connection cost by gradually deleting the minimum marginal benefit edge from the completely connected network.Finally,a numerical example is presented to demonstrate the efficiency of the algorithm.