基于局部呼叫数据建模的相对相似用户分布式查询

摘要

在众多以手机呼叫记录（ Call Detail Record ， CDR）为数据源的分析研究和挖掘应用中，相似用户查询作为基础研究方法占据着重要地位。传统的查询算法多为集中式处理，然而CDR数据分布产生和存储的本质带来了相似用户分布式查询的问题。本文结合真实数据集，分析用户存储于各基站的局部数据与全局数据的关系，提出并实现基于局部呼叫数据建模的相对相似用户分布式查询方法（ RSU-DQ），使用真实数据设计实验验证所提出方法的准确性和高效性。%In many CDR ( call detail record ) based research and data mining applications , similar user query serves as a funda-mental function .Traditional solutions are more like to be centralized query algorithms , however , the distributed nature of the pro-duction and storage of CDR posed a new problem:similar user distributed query .Based on real world data , we analyzed the rela-tionships between local data and global data , designed and implemented a local CDR-based relative similar user distributed query (RSU-DQ) approach, and then evaluated its accuracy and efficiency .