信息网络安全 ›› 2019, Vol. 19 ›› Issue (9): 36-40.doi: 10.3969/j.issn.1671-1122.2019.09.008

• • 上一篇    下一篇

基于深度学习的数据接入方法研究

谢永恒1, 冯宇波1,2, 董清风1,2, 王梅1,2   

  1. 1.北京锐安科技有限公司,北京 100192
    2.北京市网络空间数据分析与应用工程技术中心,北京 100192
  • 收稿日期:2019-07-15 出版日期:2019-09-10 发布日期:2020-05-11
  • 作者简介:

    作者简介:谢永恒(1972—),男,湖北,硕士,主要研究方向为大数据分析与挖掘;冯宇波(1984—),男,内蒙古,博士,主要研究方向为大数据及人工智能应用;董清风(1977—),男,湖北,硕士,主要研究方向为机器学习;王梅(1975—),女,辽宁,本科,主要研究方向为大数据挖掘与分析。

Research on Data Ingestion Method Based on Deep Learning

Yongheng XIE1, Yubo FENG1,2, Qingfeng DONG1,2, Mei WANG1,2   

  1. 1. Run Technologies Co., Ltd. Beijing, Beijing 100192, China
    2. Beijing Cyberspace Data Analysis and Applied Engineering Technology Research Center, Beijing 100192, China
  • Received:2019-07-15 Online:2019-09-10 Published:2020-05-11

摘要:

大数据中心在接入多源异构数据时,需要分析处理不同标准数据的标准转换问题。以往这项工作大都依赖人工分析解决。近几年快速发展的深度学习技术具有优异的特征提取和分类能力,文章基于深度学习技术设计了一套将数据形态模型和语义模型相结合的数据接入匹配算法,并利用部分样例数据进行测试验证。结果表明,该方法能够达到一定的准确率水平,并具备持续提升的空间,可以在实际应用中替代部分人工工作。

关键词: 大数据, 数据中心, 数据接入, 深度学习

Abstract:

When ingesting multi-source heterogeneous data, big data centers need to analyze standard conversion problems that deal with data of different standards. In the past, most of this work relied on manual analysis. In recent years, the fast-developing deep learning technology has excellent feature extraction and classification ability. This study designed a data access matching algorithm based on deep learning technology which combined data morphological model and semantic model, and completed verification by using sample data. The results show that the proposed method could achieve a reasonable level of accuracy, and has potential for future improvement, which could replace some manual work in many practical applications.

Key words: big data, data center, data access, deep learning

中图分类号: