Dimension structures of multidimensional model in data warehouse are hierarchical and this also leads the logical dimension structure to be hierarchical so that the dynamic characteristics of analysis, which often switches between different levels of dimension, can be better supported. Traditional algebraic lattice structure does not support more than one key in a single dimension simultaneously, while combination lattice structure supports composite keys in a single dimension by extending the traditional model. The concept of incomplete combination lattice of dimension is introduced to further improve the former combination lattice model. Incomplete combination lattice of dimension makes it possible to reduce the space of cube views while modeling data cube dimensions. Operations on combination lattice are also presented, and then used to implement an incremental method of constructing combination lattice structure.