Please wait a minute...

当期目录

    15 September 2018, Volume 4 Issue 5
    TOPIC:PRACTICAL INNOVATIONS OF BIG DATA
    An industrial big data application research using deep learning
    Guang LI, Xin YANG
    2018, 4(5):  3-14.  doi:10.11959/j.issn.2096-0271.2018046
    Asbtract ( 809 )   HTML ( 110)   PDF (1343KB) ( 1137 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    How to combine the core technologies such as big data with the smart manufacturing to increase productivity,quality and reduce costs,which is a key task for a new generation of manufacturing innovation.Aiming at the common problem of consumption redundancy to the machine tools in industry,a method based on big data and artificial intelligence to accurately predict the breakage of machine tools was proposed,which achieves an increase in the productively of machine tools and reduces the cost of production.Compared with the previous methods which use the data statistics and traditional machine learning to predict the tool wear,the spindle current value by a high speed collector was got and the strong fitting of convolutional neural network and the strong generalization ability of anomaly detection algorithms was combined.The network could get faster convergence,higher prediction accuracy and robustness.

    Analysis of online activity characteristics of hidden populations based on public data
    Chuchu LIU, Xin LU
    2018, 4(5):  15-28.  doi:10.11959/j.issn.2096-0271.2018047
    Asbtract ( 306 )   HTML ( 25)   PDF (2129KB) ( 441 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Data from 36 Baidu Tieba was collected.The sub-groups of users from three dimensions of time (temporal),content (text) and interaction (network) were analyzed,and patterns of their online activity were tried to extract and characteristics of different populations was inferred.The result indicates that the temporal pattern of posting behavior is more regular for HIV-related users.Communities followed by these users are also HIV-related,revealing a significant clustering pattern.On the other hand,MSM groups are more active during midnight,and their motivation of online activities in the community is for entertainment and for meeting partners.There is also a strong preference of following the same type of Tiebas for MSM groups.In general,HIV-related users are very concerned about their health status.On the contrary,MSM-related users are lack of awareness for protection and prevention of AIDS.

    Progress of social-big-data-based studies of music culture
    Yu GENG, Xiaopu HAN, Linyuan LÜ
    2018, 4(5):  29-40.  doi:10.11959/j.issn.2096-0271.2018048
    Asbtract ( 443 )   HTML ( 35)   PDF (1205KB) ( 637 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    In recent years,along with the popularity of the Internet and the development of technology,the rise of a series of online music communities and social websites provide a large amount of social data for the research on music culture.By combining bigdata approach and artificial intelligence methods,researchers have made a series of advances in the studies of music culture,and established a lot of new knowledge in music cultural perception,music cultural sociology,the communication of music culture,and so on.It shows the wide potential applicability of big-data-based approaches in the research of music culture.The recent social-big-data-based studies on music culture were introduced briefly,and some open problems and challenges in this issue were summarized.

    Predicting the discredited behavior of enterprises via large-scale investment network
    Tao ZHOU, Yanli LI, Qian LI, Duanbing CHEN, Wenbo XIE, Tong WU, Tu ZENG
    2018, 4(5):  41-49.  doi:10.11959/j.issn.2096-0271.2018049
    Asbtract ( 431 )   HTML ( 52)   PDF (1178KB) ( 827 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Previous enterprise credit level analysis mainly focused on the features including enterprise size,place of operation,industry category,registration and paid-in capital,and lacked in-depth analysis based on massive data.A directed investment network consisted of more than 4 million enterprises was built up,among which nearly 260 000 enterprises have various discredited behaviors.The results show that there is an obvious "network effect" in the discredited behaviors of enterprises.If the target enterprise's shareholders or its invested enterprises have discredited behaviors,the risk of having discredited behaviors of the target enterprise is far greater than the average.Based on this,a simple generalized linear regression algorithm was proposed to predict the discredited behaviors of enterprises,which is far more accurate than the regression method without considering the network effect.

    Modeling and empirical research of information redundancy on online social media
    Shuo QIN, Xin LU, Fanhui MENG, Yanqing HU
    2018, 4(5):  50-61.  doi:10.11959/j.issn.2096-0271.2018050
    Asbtract ( 486 )   HTML ( 21)   PDF (2003KB) ( 537 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    To learn more about redundancy phenomenon,a measurement to quantify this redundancy was proposed and the probability of individual redundancy in the process of information diffusion was deduced.Using simulated networks,how information propagation rate,network density,clustering coefficient affect information redundancy was investigated.The diffusion characteristics of redundant information through the actual data of Sina Weibo was observed and the perspective of information redundancy in advertising promotion and product marketing was discussed from marketing.The findings provide a new perspective for further understanding of the information diffusion characteristics of online social media.

    STUDY
    Survey on industrial big data analysis:models and algorithms
    Hongzhi WANG, Zhiyu LIANG, Jianzhong LI, Hong GAO
    2018, 4(5):  62-79.  doi:10.11959/j.issn.2096-0271.2018051
    Asbtract ( 1282 )   HTML ( 200)   PDF (998KB) ( 2653 )   Knowledge map   
    References | Related Articles | Metrics

    With the wide application of bar code,two-dimensional code,RFID,industrial sensor,automatic control system,industrial Internet,ERP and CAD/CAM/CAE techniques in industry,a large number of data related to industrial production are collected and stored in information system in real time.Analyzing those data can help to improve the production techniques,optimize the production process,reduce the production costs,laying the foundation for intelligent manufacturing.Therefore,the industrial big data analysis has drawn much attention of both industry and academia in recent years.Models and algorithms are two core issues of big data analysis theory and techniques.The concept of industrial big data analysis was introduced,and the applications of several popular models and the research results of the corresponding algorithms were reviewed,and future research directions in this area were explored.

    Stay point identification based on density
    Yurui LI, Hongmei CHEN, Lizhen WANG, Qing XIAO
    2018, 4(5):  80-93.  doi:10.11959/j.issn.2096-0271.2018052
    Asbtract ( 510 )   HTML ( 37)   PDF (1647KB) ( 739 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Identifying stay points from GPS trajectory is an important preprocessing procedure of trajectory analysis and the foundation of location based service such as user behavior analysis and personal POI recommendation,and the capability of the stay point identification method has a fundamental impact on the availability and reliability of location based service.Existing methods for identifying stay points have some shortcomings due to not considering time continuity or only considering one direction of time continuity.A new method called stay point identification based on density (SPID) was proposed.SPID takes into account the spatial-temporal clustering of trajectory points,and the time directions and time continuity of trajectory points.The experimental results on Geolife dataset verify that SPID is better than the baseline methods,and can identify two kinds of stay points which can’t be found by the baseline methods.

    APPLICATION
    DeepEye:a deep learning-based method of recognition and classification of program trading
    Guangbin XU, Wei ZHANG
    2018, 4(5):  94-102.  doi:10.11959/j.issn.2096-0271.2018053
    Asbtract ( 601 )   HTML ( 30)   PDF (1160KB) ( 968 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Program trading behavior in A-share market has been systematically analyzed based on the Shanghai Stock Exchange’s latest trading data and a feature indictor system has thus been built up for characterizing and classifying the program trading in the market.Furthermore,based on the deep learning technology,the A-share program trading intelligentized recognition and classification method,DeepEye,has been proposed,which enables program trading behavior in the market to be recognized and classified.The accuracy of the pilot implementation got about 70% which verified the feasibility and effectiveness of the new method.The proposed method can serve as an auxiliary measure to existing investor portraits and behavior supervision analysis for market regulation and can be a reference for improving the existing program trading regulatory rules.

    COLUMN:NATIONAL ENGINEERING LABORATORY FOR BIG DATA
    Big data system software eco-system and platform construction
    Jianmin WANG, Chen WANG, Yingbo LIU, Lin LIU
    2018, 4(5):  104-112.  doi:10.11959/j.issn.2096-0271.2018054
    Asbtract ( 579 )   HTML ( 61)   PDF (1720KB) ( 751 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    In view of the bottlenecks in common technology and engineering practice faced by big data system software and applications development,the key technological innovations of the National Engineering Laboratory for Big Data System Software(NEL-BDSS) were introduced systematically,including the technical architecture and domain applications of the big data system software "Tsinghua Dataway Platform",which addresses issues such as:massive multi-source heterogeneous data integration management,interactive heterogeneous data analysis framework,data visualisation and intelligent data engineering,validation and verification of hybrid source big data software,and domain-specific big data applications development and run-time environment.The big data system software eco-system,state-of-the-art big data technology and systems,domain applications,as well as future challenges were summarized systematically.The NEL-BDSS focuses on supporting demonstrative applications of industrial big data,environmental big data as well as meteorological big data.

Most Download
Most Read
Most Cited