基于约束的多维数据挖掘技术

Constraint based multidimensional data mining technology

从特定查询驱动的系统功能出发 ,讨论了面向多维数据挖掘的重点——五种约束 ,通过一个数据挖掘查询实例 ,进一步阐述了这些约束及所产生的关联规则 ,并用数据挖掘查询语言 (DMQL )进行表达。介绍和讨论了关联规则的处理 ,在挖掘关联规则中 ,联合使用维 /层约束和规则约束 ,能够带来高效的挖掘过程 ,一个规则约束如果能被较深地推入到分层结构里 ,进一步在当前提取层和较深层上挖掘 ,是非常有价值的。最后给出了运用本文思想的一种联机分析挖掘系统的结构 ,并对其组成。

Based on the functions of ad hoc and query driven system, the essentials five constraints for multidimension data mining are discussed. The constraints and the association rules are interpreted using an example of data mining query, and it is expressed with DMQL. Association rules handling is discussed. It produces efficient data mining that dimension/level and rule constraints are used together in mining association rules. It is interesting whether a rule constraint can be pushed deeply into the hierarchy to facilitate mining at the current level of abstraction and deeper levels. In the end, this paper provides a OLAM architecture, its composition, function and benefits are presented.