Abstract: In recent years,multigranulation approach is a new arisen data analyzing model. To deal with the incomplete information system through the multigranulation approach, by employing the basic ideas of non-symmetric similarity relation and variable precision, the multigranulation based variable precision rough sets were proposed, which include optimistic and pessimistic cases, respectively. Not only the basic properties about these models were proved,but also the relationships between variable precision approach and strict inclusion approach were discussed under the multigranulation frame. Finally,an example was used to show the "OR" decision rules can be generated from the incomplete information systems by the proposed multigranulation based variable precision rough set models.

Key words: Multigranulation rough sets, Variable precision rough set, Incomplete information system, Similarity relation