首页> 外国专利> SPAM FILTERING MODEL LEARNING METHOD FOR FILTERING SHORT SPAM MESSAGE, METHOD AND APPARATUS FOR FILTERING SHORT SPAM MESSAGE USING THE SAME

SPAM FILTERING MODEL LEARNING METHOD FOR FILTERING SHORT SPAM MESSAGE, METHOD AND APPARATUS FOR FILTERING SHORT SPAM MESSAGE USING THE SAME

机译:用于过滤短垃圾邮件的垃圾邮件过滤模型学习方法,使用该方法过滤短垃圾邮件的方法和装置

摘要

The invention and SMS spam filtering model training method using the same spam message filtering for spam filtering of SMS messages It relates to a method and spam short message filtering unit. Spam filtering model learning process according to the invention (a) a plurality of spam messages, and a plurality of non-spam learning corpus which is a step input is classified as a message; (B) extracting the stylistic qualities of qualities based on vocabulary and language represented a form based on the information from the learning corpus; (C) a step of selecting a plurality of training qualities based on the information gain of the extracted vocabulary qualities and qualities of the extracted style (Information gain value); (D) characterized by applying a statistical classification model based on the plurality of training quality by including the step of creating a spam filtering model. In this way, as well as the contents of a short message, such as SMS messages, whether the content is created in any way, or any language that an expression is created to reflect the form, to significantly reduce the error of classifying a non-spam messages to a spam message can.
机译:一种垃圾邮件过滤方法和短信垃圾邮件过滤模型训练方法相同的垃圾邮件过滤方法,涉及一种垃圾短信过滤方法和单元。根据本发明的垃圾邮件过滤模型学习过程(a)将多个垃圾邮件和作为步骤输入的多个非垃圾邮件学习语料分类为邮件; (B)基于词汇和语言提取质量的风格特征,该特征是基于来自学习语料库的信息的形式; (C)基于所提取的词汇质量的信息增益和所提取的风格的质量(信息增益值)来选择多个训练质量的步骤; (D)的特征在于,通过包括创建垃圾邮件过滤模型的步骤,基于多个训练质量应用统计分类模型。以此方式,以及短消息(例如SMS消息)的内容,无论是以任何方式创建内容,还是以创建表达该表格的形式表达的任何语言,都可以大大减少对非短信内容进行分类的错误-spam邮件可以转换为垃圾邮件。

著录项

  • 公开/公告号KR101104602B1

    专利类型

  • 公开/公告日2012-01-12

    原文格式PDF

  • 申请/专利权人

    申请/专利号KR20090118504

  • 发明设计人 손대능;임해창;이정태;

    申请日2009-12-02

  • 分类号H04W4/14;H04W48/02;

  • 国家 KR

  • 入库时间 2022-08-21 17:08:43

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号