| 本文已被：浏览 4236次 下载 5567次 |
码上扫一扫！ |
|
|
| 一种无线通信环境中用户移动模式的挖掘算法 |
|
宋国杰1, 唐世渭1,2, 杨冬青1, 王腾蛟1, 叶恒强3
|
|
1.北京大学,计算机科学技术系,北京,100871;2.北京大学,视觉与听觉信息处理国家重点试验室,北京,100871;3.广州新太科技股份有限公司,广东,广州,510665
|
|
| 摘要: |
| 发现无线通信环境中用户的移动模式是移动对象管理中的一个关键问题.提出一种快速挖掘该模式的算法SAM(split and merge),用来挖掘移动对象所产生有序数据集中潜在的移动模式,从而为移动对象管理提供服务.该算法将自底向上搜索和自顶向下过滤技术相结合,采用图存储压缩数据集方法,利用非频繁项集分解子图和频繁长模式过滤数据集相结合的技术,大大减少了迭代次数,降低了CPU时间.最后给出了算法性能比较和算法分析.结果表明,该算法是有效的. |
| 关键词: 数据挖掘 最大频繁项集 移动模式 移动对象管理 移动通信 |
| DOI： |
| 分类号: |
| 基金项目:国家重点基础研究发展规划973资助项目(G1999032705);北京大学-IBM创新研究院资助项目 |
|
| An Algorithm of Mining Personal Moving Patterns in a Wireless Communication Environment |
|
SONG Guo-jie,TANG Shi-wei,YANG Dong-qing,WANG Teng-jiao,YE Heng-qiang
|
| Abstract: |
| Discovering moving pattern is a key problem of mobile management in wireless communication. In this paper, an algorithm named SAM (split and merge) is proposed to mine MFMP in sequential datasets of moving object, and then to provide services for moving object management. This algorithm combines the bottom-up search and top down filter and uses data structure——graph to store datasets, infrequent item sets to split moving pattern graph, and long moving pattern to filter data sets strategy, and then the iteration number and CPU time are reduced greatly.Lastly,the performance analysis and the comparison of the algorithms are provided.Experimental results show that the SAM algorithm outperfprms other existing algorithms. |
| Key words: data mining long pattern moving pattern mobile object management mobile communication |
