基于BETR-BiGRU-CRF模型的文物档案知识图谱构建研究——以北京中轴线文化遗产档案为例

建立文物档案知识图谱便于对文物档案进一步的研究与利用。以中轴线文化遗产档案为原始数据集,采用BERT-BiGRU-CRF模型进行文物档案实体识别,在此基础上构建中轴线文化遗产档案知识图谱,使用Neo4j图数据库完成知识存储。经实验验证,BERT-BiGRU-CRF模型在文物档案实体识别任务中具有更高的精确度与召回率,有助于建立准确、一致、完整的文物档案知识图谱,能够为文物档案管理与保护、文化遗产保护与传承等提供参考与借鉴。

An Innovative K-Anonymity Privacy-Preserving Algorithm to Improve Data Availability in the Context of Big Data

The development of technologies such as big data and blockchain has brought convenience to life,but at the same time,privacy and security issues are becoming more and more prominent.The K-anonymity algorithm is an effective and low computational complexity privacy-preserving algorithm that can safeguard users’privacy by anonymizing big data.However,the algorithm currently suffers from the problem of focusing only on improving user privacy while ignoring data availability.In addition,ignoring the impact of quasi-identified attributes on sensitive attributes causes the usability of the processed data on statistical analysis to be reduced.Based on this,we propose a new K-anonymity algorithm to solve the privacy security problem in the context of big data,while guaranteeing improved data usability.Specifically,we construct a new information loss function based on the information quantity theory.Considering that different quasi-identification attributes have different impacts on sensitive attributes,we set weights for each quasi-identification attribute when designing the information loss function.In addition,to reduce information loss,we improve K-anonymity in two ways.First,we make the loss of information smaller than in the original table while guaranteeing privacy based on common artificial intelligence algorithms,i.e.,greedy algorithm and 2-means clustering algorithm.In addition,we improve the 2-means clustering algorithm by designing a mean-center method to select the initial center of mass.Meanwhile,we design the K-anonymity algorithm of this scheme based on the constructed information loss function,the improved 2-means clustering algorithm,and the greedy algorithm,which reduces the information loss.Finally,we experimentally demonstrate the effectiveness of the algorithm in improving the effect of 2-means clustering and reducing information loss.

Research on Tensor Multi-Clustering Distributed Incremental Updating Method for Big Data

The scale and complexity of big data are growing continuously,posing severe challenges to traditional data processing methods,especially in the field of clustering analysis.To address this issue,this paper introduces a new method named Big Data Tensor Multi-Cluster Distributed Incremental Update(BDTMCDIncreUpdate),which combines distributed computing,storage technology,and incremental update techniques to provide an efficient and effective means for clustering analysis.Firstly,the original dataset is divided into multiple subblocks,and distributed computing resources are utilized to process the sub-blocks in parallel,enhancing efficiency.Then,initial clustering is performed on each sub-block using tensor-based multi-clustering techniques to obtain preliminary results.When new data arrives,incremental update technology is employed to update the core tensor and factor matrix,ensuring that the clustering model can adapt to changes in data.Finally,by combining the updated core tensor and factor matrix with historical computational results,refined clustering results are obtained,achieving real-time adaptation to dynamic data.Through experimental simulation on the Aminer dataset,the BDTMCDIncreUpdate method has demonstrated outstanding performance in terms of accuracy(ACC)and normalized mutual information(NMI)metrics,achieving an accuracy rate of 90%and an NMI score of 0.85,which outperforms existing methods such as TClusInitUpdate and TKLClusUpdate in most scenarios.Therefore,the BDTMCDIncreUpdate method offers an innovative solution to the field of big data analysis,integrating distributed computing,incremental updates,and tensor-based multi-clustering techniques.It not only improves the efficiency and scalability in processing large-scale high-dimensional datasets but also has been validated for its effectiveness and accuracy through experiments.This method shows great potential in real-world applications where dynamic data growth is common,and it is of significant importance for advancing the development of data analysis technology.

基于高光谱成像和Att-BiGRU-RNN的柑橘病叶分类

为实现对柑橘叶片病虫药害种类的快速精准识别,针对多种类柑橘病叶设计一种融合注意力机制(Attention mechanism)的双向门控循环单元-循环神经网络(Attention-bidirectional gate recurrent unit-recurrent nural network, Att-BiGRU-RNN)分类模型。该模型在编解码模块分别采用BiGRU和RNN结构,能够利用高光谱数据前后波段光谱信息的关联性,有效提取光谱信息的深层特征;根据不同波段光谱信息的差异性引入注意力机制动态分配权重信息,提高重要光谱特征对分类模型的贡献率,提升模型的分类准确率。获取6类柑橘叶片高光谱信息,构建实验样本集,利用Att-BiGRU-RNN、VGG16、SVM和XGBoost分别建立柑橘病叶分类模型,Att-BiGRU-RNN模型总体分类准确率(Overall accuracy, OA)平均可达98.21%,相较于其他3种模型分别提高4.71、10.95、3.89个百分点,对光谱曲线重合度高的除草剂危害和煤烟病叶片的分类准确率有显著提升。实验结果表明,深度学习方法可有效利用高光谱不同波段间的关联信息,识别准确率较机器学习方法有大幅提高,为柑橘病虫药害快速无损检测和防治提供了一种新方法。

融合深度BiGRU与全局图卷积的方面级情感分析模型

现有基于深度学习的方面级情感分析模型需要考虑如何提取深层次的语义信息,其次通过依存树提取句法结构时可能存在信息丢失与数据稀疏问题.针对以上问题,本文提出了基于深度双向门控循环单元与全局双向图卷积网络的神经网络模型(DBG-GBGCN).该模型通过深度双向门控循环单元捕获深层次的语义特征,得到上下文的隐层表示.然后将依存树的邻接矩阵转变为带有全局句法信息的全局矩阵,将此矩阵与上下文的隐层表示一起输入至双向图卷积网络进行特征融合,最后经过掩码层和注意力层得到一个包含深层语义特征与句法结构信息结合的分类特征.实验结果证明,该模型在5个公开数据集上的准确率与F1值均比对比模型有着一定的提升.

Overview of the identification of traffic accident-prone locations driven by big data

Effective identification of traffic accident-prone points can reduce accident risks and eliminate safety hazards.This paper first systematically compares the research in Chinese and foreign literature,and proposes three types of identification indicators,namely absolute,relative and comprehensive,according to different reference standards.According to the evaluation indicators and modelling methods,the current status of research and problems in identification theory and methods are systematically summarised in terms of mathematical statistics,cluster analysis,machine learning and conflict technology.The study shows that the foreign literature focuses on the innovation of data and indicators and changes from accident point safety management to road network safety management,while the research in Chinese literature focuses on the integration of multiple identification methods and theoretical innovation.Driven by big data,the identification of traffic accident-prone points has been further developed at the meso-micro scale.Morphological image processing methods are widely used,combined with GIS platforms,to accurately mine the spatial attributes and correlations of accidents.Also,considering the spatial and temporal distribution of accidents,the identification results are also transformed from regions to specific road sections and points to achieve more accurate identification.

基于CNN-BiGRU的复杂连续人体活动Wi-Fi感知方法

基于Wi-Fi信道状态信息(CSI,channel state information)的人体活动感知在虚拟现实、智能游戏、元宇宙等未来智能交互场景具有重要的应用前景,复杂连续人体活动的精准感知是Wi-Fi感知的重要挑战。卷积神经网络(CNN,convolutional neural network)具备空间特征提取能力,但对数据的时序特征建模能力差。而适用于时间序列数据建模的长短期记忆(LSTM,long short-term memory)网络或门控循环单元(GRU,gated recurrent unit)网络忽视了对数据空间特征的学习。针对此问题,提出了一种融合双向门控循环单元(BiGRU,bidirectional gated recurrent unit)网络的改进型CNN。所提网络利用BiGRU的双向特征提取能力捕捉时序数据前后信息的关联和依赖性,实现时序CSI数据的时空特征提取,进而呈现动作与CSI数据的映射关系,从而提高对复杂连续动作的识别精度。以篮球动作为场景对所提网络结构进行了实验,结果表明,该方法在多种条件下识别准确率均高于95%,与传统多层感知机(MLP,multi-layer perceptron)、CNN、LSTM、GRU、具有注意力机制的双向长短期记忆(ABLSTM,attention based bidirectional long short-term memory)网络等基线方法相比,识别准确率提升了1%~20%。

基于BiGRU和贝叶斯分类器的文本分类

针对传统的循环神经网络模型在处理长期依赖问题时面临着梯度爆炸或者梯度消失的问题,且参数多训练模型时间长,提出一种基于双向GRU神经网络和贝叶斯分类器的文本分类方法。利用双向GRU神经网络提取文本特征,通过TF-IDF算法权重赋值,采用贝叶斯分类器判别分类,改进单向GRU对后文依赖性不足的缺点,减少参数,缩短模型的训练时间,提高文本分类效率。在两类文本数据上进行对比仿真实验,实验结果表明,该分类算法与传统的循环神经网络相比能够有效提高文本分类的效率和准确率。

Privacy-preserving deep learning techniques for wearable sensor-based big data applications

Wearable technologies have the potential to become a valuable influence on human daily life where they may enable observing the world in new ways,including,for example,using augmented reality(AR)applications.Wearable technology uses electronic devices that may be carried as accessories,clothes,or even embedded in the user's body.Although the potential benefits of smart wearables are numerous,their extensive and continual usage creates several privacy concerns and tricky information security challenges.In this paper,we present a comprehensive survey of recent privacy-preserving big data analytics applications based on wearable sensors.We highlight the fundamental features of security and privacy for wearable device applications.Then,we examine the utilization of deep learning algorithms with cryptography and determine their usability for wearable sensors.We also present a case study on privacy-preserving machine learning techniques.Herein,we theoretically and empirically evaluate the privacy-preserving deep learning framework's performance.We explain the implementation details of a case study of a secure prediction service using the convolutional neural network(CNN)model and the Cheon-Kim-Kim-Song(CHKS)homomorphic encryption algorithm.Finally,we explore the obstacles and gaps in the deployment of practical real-world applications.Following a comprehensive overview,we identify the most important obstacles that must be overcome and discuss some interesting future research directions.

国家建筑博物馆BIG迷宫,华盛顿,美国

一览无余是迷宫?一座迷宫往往越是深入,路线就越曲折。假如我们将这种套路反转过来,创造一座让人在达到迷宫中心时豁然开朗、一目了然的龙门阵会怎样?从外面看,迷宫的立方体造型将最终的观赏点隐藏在约5.5m(18ft)高的墙之后。从里面看,墙体缓慢向中心降低,以开阔的观赏点终结——让人360°环视自己的来龙去脉。

基于BiGCN和IAM的方面级情感分类模型

目前基于神经网络的方面级情感分类模型很少会考虑上下文单词与方面词之间的句法依存关系,可能会错误地将与方面词语法无关的上下文单词作为方面词的情感特征;另一方面大多数方法也忽略了上下文与方面词之间的交互信息。针对这两个问题,提出了基于双向图卷积网络(BiGCN)和交互注意力机制(IAM)的方面级情感分类模型(BiGCN-IAM),该模型在句法依存树上使用双向图卷积网络提取上下文单词和方面词之间的句法依存关系,然后使用掩码层得到特定的方面词表示;最后使用交互注意力机制学习上下文与方面词之间的交互信息,同时提取了上下文中的重要情感特征和方面词中对分类有贡献的特征。通过在五个公开数据集上的实验证明,该模型效果优于基线模型。

基于VDCNN和BiGRU混合的文本分类研究

卷积神经网络(CNN)及循环神经网络(RNN)在自然语言方面存在着广泛的应用,但仅依靠CNN无法有效的处理自然语言中的上下文信息,RNN则在应用过程中常会出现梯度消失、梯度爆炸的现象,从而限制了文本分类的准确率。基于此,构建了基于超深卷积神经网络(VDCNN)和双向门控循环(BiGRU)神经网络的混合模型。模型首先利用VDCNN的进行文本向量局部特征的提取,利用BiGRU提取与上下文信息相关的全局特征;然后将提取到的局部特征与全局特征进行融合;最后将融合后的特征信息放入softmax函数实现对于文本内容的分类。利用20Newsgroups和MR数据集对VDCNN-BiGRU模型进行验证,通过与VDCNN、GRU、BiGRU三个模型的对比,结果表明VDCNN-BiGRU模型能够有效提升文本分类的精度。

An Ui Design Optimization Strategy for General App in Big Data Environment

Due to the huge amount of increasing data, the requirements of people forelectronic products such as mobile phones, tablets, and notebooks are constantlyimproving. The development and design of various software applications attach greatimportance to users’ experiences. The rationalized UI design should allow a user not onlyenjoy the visual design experience of the new product but also operating it morepleasingly. This process is to enhance the attractiveness and performance of the newproduct and thus to promote the active usage and consuming conduct of users. In thispaper, an UI design optimization strategy for general APP in the big data environment isproposed to get better user experience while effectively obtaining information. Anexperimental example of a library APP is designed to optimize the user experience. Theexperimental results show that the user-centered UI design is the core of optimization,and user portrait based on big data platforms is the key to UI design.

The Analysis of China’s Integrity Situation Based on Big Data

In order to study deeply the prominent problems faced by China’s clean government work,and put forward effective coping strategies,this article analyzes the network information of anti-corruption related news events,which is based on big data technology.In this study,we take the news report from the website of the Communist Party of China(CPC)Central Commission for Discipline Inspection(CCDI)as the source of data.Firstly,the obtained text data is converted to word segmentation and stop words under preprocessing,and then the pre-processed data is improved by vectorization and text clustering,finally,after text clustering,the key words of clean government work is derived from visualization analysis.According to the results of this study,it shows that China’s clean government work should focus on‘the four forms of decadence’issue,and related departments must strictly crack down five categories of phenomena,such as“illegal payment of subsidies or benefits,illegal delivery of gifts and cash gift,illegal use of official vehicles,banquets using public funds,extravagant wedding ceremonies and funeral”.The results of this study are consistent with the official data released by the CCDI’s website,which also suggests that the method is feasible and effective.

Environmental Information Governance Reform in the Era of Big Data--From Information Disclosure to Public Service

The development of environmental information governance includes three phases: providing for oneself,information disclosure,and public service. And then China is in the transition and transformation of environmental information disclosure to the environmental information public service. The core of the transformation is public participation,in the whole procedure of environmental information supply decision making,production,and quality supervision and evaluation,etc. The target path of the environmental information governance reform includes five parts: improvement of public satisfaction,optimizing information disclosure,information quality control,integration of information resources,and multiple supply.

基于BiGRU-Capsule的多标签文本分类

传统的文本分类一般采用单标签形式,但现实生活中多标签文本比单标签文本具有更广泛的应用场景。本文提出一种BiGRU-Capsule模型的多标签文本分类方法,该方法首先通过嵌入层将输入的文本序列转化为向量表示;然后通过BiGRU和Capsule提取文本特征;最后使用sigmoid分类器进行分类。为确保数据量足够,利用今日头条2018新闻标题多标签语料数据集进行实验,将胶囊网络模型作为对比模型进行多标签文本分类实验与分析。实验结果表明:本文模型的多标签文本分类效果得到有效提升。

Research on Teaching Innovation of Property Insurance Course:Based on the Perspective of Big Data Development

The development of big data has brought unprecedented challenges and opportunities to the teaching reform of higher education.Property insurance course is the core course of economics and management,and it is the guarantee for the supply of talents in the health financial market.Big data technology and data economy put forward innovative requirements for its teaching objectives,teaching content,and teaching system.In China’s new round of double-first-class universities and disciplines,big data is an important foundation and driving force.The comprehensive integration of property insurance and big data is reflected in:Cultivate students’big data thinking;Cultivate students’practical application ability based on market employment needs;Build a new discipline system of applied economics,and achieve good coordination between property insurance courses and other disciplines;The government,enterprises and universities form a strategic partnership to jointly participate in the development and construction of courses;The formulation of government policies can have a better governance effect on the development of higher education and talent training.

Mapping Research Fronts of Big Data Study in 21st Century

In this paper, we aim at researching status and finding the research fronts in the field of Big Data in recent 15 years. This study applies CiteSpace software to the visualization analysis about the literature data from the year 2000 to 2014 in Web of Science, and to carry out preliminary discussion on the research hotspot and the fronts of Big Data. It concluded that Big Data have been exerting so great influence on the world that increasing researchers from different countries and institutes have continuously researched it in recent years.

Big Data Research in Italy： A Perspective

The aim of this article is to synthetically describe the research projects that a selection of Italian univer- sities is undertaking in the context of big data. Far from being exhaustive, this article has the objective of offering a sample of distinct applications that address the issue of managing huge amounts of data in Italy, collected in relation to diverse domains.

Social Media and Stock Market Prediction: A Big Data Approach

Big data is the collection of large datasets from traditional and digital sources to identify trends and patterns.The quantity and variety of computer data are growing exponentially for many reasons.For example,retailers are building vast databases of customer sales activity.Organizations are working on logistics financial services,and public social media are sharing a vast quantity of sentiments related to sales price and products.Challenges of big data include volume and variety in both structured and unstructured data.In this paper,we implemented several machine learning models through Spark MLlib using PySpark,which is scalable,fast,easily integrated with other tools,and has better performance than the traditional models.We studied the stocks of 10 top companies,whose data include historical stock prices,with MLlib models such as linear regression,generalized linear regression,random forest,and decision tree.We implemented naive Bayes and logistic regression classification models.Experimental results suggest that linear regression,random forest,and generalized linear regression provide an accuracy of 80%-98%.The experimental results of the decision tree did not well predict share price movements in the stock market.