基于多维数据模型的交叉层关联规则挖掘

多层关联规则是带有一定概念分层的关联规更哇，它描述了不同抽象级别上数据项之间的关联性，且不同级别上的关联性具有不同的指导意义．但目前已讨论的多层关联规则，大都局限于挖掘同一抽象层上数据项之间的关联，因而，针对这一问题，本文对已有的FP—Tree算法进行扩充和改进，实现了既能挖掘同一抽象层上也能挖掘不同抽象层上数据项之间关联性的多层关联挖掘算法，即交叉层关联规则挖掘算法FP—Tree＊．同时，在算法实施之前，还结合多层关联挖掘本身的特点，对现有的数据存储结构进行改进，提出用字符序列对事务项编码的方法，从而简化了大量的数据预处理工作．

Mining Cross-Level Association Rules Based on Multi-Dimensional Data Model

Multi-level association rule is the rule with the characteristics of concept hierarchies. It describes the associations among the data items at different abstract levels, and the associations at different levels have different meaning of guidance. However, all multi-level association rules under discussion at present are limited to mining associations among data items at the same level. Aimed at this issue, this paper gives expansion and refinement to the already existing FP-Tree algorithm, and actualizes the FP_-Tree* algorithm, which has the function of mining association rules among data items both at the same and different abstract levels, namely, cross-level association rules mining. Moreover, before actualizing the algorithm, this paper refines the existing data storage schema and presents a method of encoding the items as a sequence of letters, in combination with its characteristics, so as to simplify the mass data preprocessing.