计算机与现代化

• 数据库与数据挖掘 • 上一篇    下一篇

基于决策树的多维属性自动推理识别

  

  1. (1.武汉邮电科学研究院,湖北武汉430074;2.烽火通信科技股份有限公司南京研发部,江苏南京210019)
  • 收稿日期:2016-06-28 出版日期:2017-03-09 发布日期:2017-03-20
  • 作者简介:汤鲲(1979-),男,江苏南京人,武汉邮电科学研究院和烽火通信科技股份有限公司南京研发部高级工程师,硕士,研究方向:网络安全; 蒋炳南(1989-),男,湖北宜昌人,硕士研究生,研究方向:大数据分析; 彭艳兵(1974-),男,高级工程师,博士,研究方向:网络行为分析,海量数据挖掘。

Automatic Reasoning Identification of Multidimensional Attribute Based on Decision Tree

  1. (1. Wuhan Research Institute of Posts and Telecommunications, Wuhan 430074, China;

    2. Nanjing Research and Development Department, FiberHome Communication Technology Co. Ltd., Nanjing 210019, China)
  • Received:2016-06-28 Online:2017-03-09 Published:2017-03-20

摘要:

现有的决策树分类在属性识别的应用中存在一定的不足,如样本数据属性类别必须事先已知,无法做到自动推理等。针对以上不足,本文提出一种基于决策树的多维属性自动推理的机器学习识别模型。通过引入属性相似度度量策略和机器学习的方法,实现多维属性的自动推理和识别。实验结果表明,该模型能有效地对多维属性进行自动分类,准确率达到93%左右,且识别的最终得分score均在0.71以上,能很好地满足属性自动识别的需求。

关键词: 决策树分类, 多维属性, 自动推理, 相似度度量策略, 机器学习

Abstract:

There exist certain disadvantages to the decision tree classification in the application of attribute recognition, such as sample data attribute category must be known in advance, can’t do automatic reasoning and so on. Against the above, this paper proposes an automatic reasoning machine learning identification model of multidimensional attribute based on decision tree. The paper introduces attribute similarity measure strategy and machine learning method to realize automatic reasoning and recognition for multi-attribute. The experimental results show that this model can effectively realize automatic classification of multi-attribute, accuracy is about 93%, and the final score of identification is above 0.71, which can well meet the requirements of automatic identification of attributes.

Key words: decision tree classification, multi-attribute, automatic reasoning, similarity measure, machine learning

中图分类号: