The Dimensional Fact Model is a conceptual model that allows to design the multidimensional schema of a data warehouse. Its methodology is based on the (re)modelling of the schema of a relational database; such a schema is represented by a tree, and the (re)modelling consists of the traditional operations on graphs (as prune, graft, adding child, change parent,⋯). The tree is generated automatically by an algorithm, starting from the logical schema of a relational database. Nowadays, this algorithm is not able to manage many-to-many relationships, and then, it stops the tree creation when it meets this kind of relationships. In this work, we propose a novel methodology, able to overcome this algorithmic limit, via the declarative approach of the logical programming.