面向政策法规数据的分类方法  

A Categorization Method for the Policies and Regulations Data

作　　者：苏变萍[1] 侯筱婷[1] 

机构地区：[1]西安建筑科技大学理学院,陕西西安710054

出　　处：《微电子学与计算机》2008年第7期166-168,172,共4页Microelectronics & Computer

基　　金：陕西省自然科学基金项目(2004G05);建设部计划项目(02-5-1.65)

摘　　要：通过设计阀值、构造类别的基向量和待分类向量的复数指标,建立了面向政策法规数据的适于有效机器学习的实时动态可扩展的分类方法.改进了文档的自动分类多采用以类别为中心的分类模式,综合了以文档为中心和以类别为中心两种分类模式的优点,实现了一次性完成单标号分类和多标号分类问题,克服了以往分类中预先定义主题类别和聚类分类中预先指定类别的种数的缺陷.编制了相应的算法.Facing policy and regulations data, by means of the designed threshold, the complex indicators and the structure-based vector to be categorized, it established a real-time extensible categorization method suitable for dynamic machine learning. This method improved the former automatic documentation categorization model which basing on categorypivoted categorization and combines the advantages of both category-pivoted categorization and document-pivoted categorization. Solves the problems in single-label categorization and multilabel categorization and overcomes the former shortcomings that it should pre-defined categories theme in the categorization and pre-specified categories number in the duster categorization. Finally it established the corresponding algorithm.

关 键 词：文本挖掘 分类 模型 算法 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]