K-Means Based Approach for OLAP Dimension Updates

Actual data warehouses models usually consider OLAP dimensions as static entities. However, in practice, structural changes of dimensions schema are often necessary to adapt the multidimensional database to changing requirements. This paper presents a new structural update operator for OLAP dimensions. This operator can create a new level to which, a pre-existent level in an OLAP dimension hierarchy rolls up. To define the domain of the new level and the aggregation function from an existing level to the new level, our operator classifies all instances of an existing level into k clusters with the k-means clustering algorithm. To choose features for k-means clustering, we propose two solutions. The first solution uses descriptors of the pre-existent level in its dimension table while the second one proposes to describe the new level by measures attributes in the fact table. Moreover, we carried out some experimentations within Oracle 10 g DBMS which validated the relevance of our approach.