Dealing with multi-valued data has become quite common in both the framework of databases as well as data analysis. Such data can be constrained by domain knowledge provided by relations between the variables and these relations are expressed by rules. However, such knowledge can introduce a combinatorial increase in the computation time depending on the number of rules. In this paper, we present a way to cluster such data in polynomial time. The method is based on the following: a decomposition of the data according to the rules, a suitable dissimilarity function and a clustering algorithm based on dissimilarities.
