Abstract: In frequent neighboring class sets mining, for the dynamic change of constraint condition given by user, present mining algorithms have redundant computing since they over and over scan spatial transactions. Hence, this paper proposes a dynamic constraint mining algorithm of frequent neighboring class sets, which can extract frequent neighboring class sets meeting user demand according to dynamic constraint instruction given by user. The algorithm uses array index to map neighboring class sets, and uses positive integer power set method to compute support and search frequent neighboring class sets meeting user’s dynamic constraint. The algorithm needn’t create candidate frequent neighboring class sets and repeat scanning spatial transaction from buffer analysis. In order to verify the algorithm is practical and effective, it is applied to mobile environment to shorten response time of mobile system to try their best to improve client satisfaction. In mobile computing the simulate experiment indicates that the algorithm is faster and more efficient than present algorithm.

Key words: neighboring class sets, dynamic constraint, positive integer power set method, mobile computing, spatial data mining

摘要： 在频繁邻近类别集挖掘中，由于用户指定约束条件的动态变化，现有挖掘算法因多次重复扫描空间事务而存在冗余计算，故提出一种频繁邻近类别集的动态约束挖掘算法，其能根据用户发出的动态约束指令，提取满足用户需求的频繁邻近类别集；该算法用数组索引映射邻近类别集，用正整数幂集法计算支持数和搜索满足用户动态约束的频繁邻近类别集；该算法无需产生候选频繁邻近类别集且不重复扫描缓冲分析得到的空间事务；为了验证算法的实用性和高效性，将其应用到移动环境中缩短移动系统的响应时间，尽最大努力来提高用户满意度，通过移动计算下的仿真实验表明该算法比现有算法更快速更有效。

关键词: 邻近类别集, 动态约束, 正整数幂集法, 移动计算, 空间数据挖掘