分维度策略的异网宽带用户识别方案  

A fractional dimension strategy for user identification of different broadband

作　　者：王冰莹[1,2] 王晶[1,2] 

机构地区：[1]杭州东信北邮信息技术有限公司 [2]北京邮电大学网络与交换技术国家重点实验室

出　　处：《电信网技术》2016年第11期73-76,共4页Telecommunications Network Technology

摘　　要：目前某运营商已有的异网宽带识别模型采用用户使用异网IP登陆自有网站这种单一维度分析的方式,识别准确率较低且覆盖用户数过少。采用多维度分析的方式构建模型可以提高准确率和效率,但是不同维度的数据量级差异明显,如果强行建立模型分析会导致出现大量的空值。针对这些问题,本文设计了一种分维度策略的异网宽带用户识别方案,实际应用表明,该设计行之有效。Currently, some operators already adopt a method for user identification of different broadband. The method is a single dimensional analysis, that is, analysis the behavior of other IP landing operator＇ s own web sites. But the identification correct rate and user coverage rate is low. Adopting multi-dimensional analysis method formulates the model could improve its accuracy and efficiency. However, different dimensional data gap obvious. If forced to establish the model, it might cause many vacant data. Aiming at these problems, this paper designed a fractional dimension strategy plan and the practice shows its effectiveness.

关 键 词：分维度策略 异网宽带 模型 

分 类 号：TP393.0[自动化与计算机技术—计算机应用技术]