电信科学 ›› 2016, Vol. 32 ›› Issue (5): 160-165.doi: 10.11959/j.issn.1000-0801.2016153

• 运营技术广角 • 上一篇    下一篇

基于预测度量值的IPTV用户行为规则预测算法

汪敏娟1,2,嵇正鹏2,吕超1,2   

  1. 1 江苏省公用信息有限公司,江苏 南京 210006
    2 中国电信股份有限公司智慧家庭运营中心,江苏 南京 210006
  • 出版日期:2017-02-22 发布日期:2017-02-22

A rules extraction algorithm for IPTV customers forecasting based on the forecasting entropy measurement

Minjuan WANG1,2,Zhengpeng JI2,Chao LV1,2   

  1. 1 Jiangsu Public Information Co.,Ltd.,Nanjing 210006,China
    2 Smart Home Operation Center of China Telecom Co.,Ltd.,Nanjing 210006,China
  • Online:2017-02-22 Published:2017-02-22

摘要:

提出了一种符合用户行为的,基于海量IPTV用户特征数据,对IPTV用户进行分群和规则提取的算法模型。首先提出了符合用户点播使用行为的IPTV用户分群的描述维度,即通过基础属性描述用户分群、通过点播行为描述用户分群变化趋势。然后提出了预测度量值的概念,对用户分群的稳定性进行描述,并提出了对稳定的用户分群提取点播行为概率的算法。最后通过大量的IPTV运营数据对算法模型进行了验证分析。

关键词: IPTV, 点播行为, 等价类划分, 信息熵, 预测度量值, 规则提取

Abstract:

An algorithm model conformed to the user behavior,based on the massive IPTV user characteristic data which extract rules and classify IPTV users was proposed.First,IPTV user group description dimension in accordance with the user on demand was put forward.Namely,the user group could be described by basic property and trend of user behavior could be described by users' demand behavior.Then the concept of prediction measurement was put forward,the stability of user group was described,and an algorithm which extracted demand behavior probability on stable user group was proposed.At last,the algorithm model was verified and analyzed by massive IPTV operation data.

Key words: IPTV, demand behavior, equivalent class, information entropy, prediction measurement, rule extraction

No Suggested Reading articles found!