1) f: X---->Y is a dependency relation 2) s1 is a similarity relation in the domain X 3) s2 is a similarity relation in the domain Y 4) suppose f(x1)=y1 and f(x2)=y2, and x1 and x2 are similar with respect to s1, then y1 and y2 are also similar with respect to s2.A similarity relation s1 is an equivalence relation and induces an equivalence partitioning X/s1 on the domain X. Similarly, a similarity relation s2 is an equivalence relation and induces an equivalence partitioning Y/s2 on the domain Y. What we are saying in condition 4) is that the homological image of X/s1 is a refinement of Y/s2. In other words, if X1 of X is an equivalence set and f(X1) its homological image, then f(X1) must be contained in some equivalence set Y2 of Y. In fact, this procedure can be used to effectively determine wither f is a type-M dependency relation, when the cardinality of the equvalence partitioning is finite for both similarity relations.
From the above definition it is clear that X---I,s2--->Y is a type-M dependency relation where I is the identity relation and s2 any similarity relation. In particular, X---I,I--->Y is a type-M dependency relation. In other words if we use identity relations as similarity relations, we can regard any dependency relation as a type-M dependency relation. Therefore the type-M normal forms will be identical to the usual normal forms, as long as we use identity relations as similarity relations.
The above discussion shows type-M dependency relations are specializations of the usual dependency relations and therefore we are on safe ground. However type-M dependency relations are more strict than the usual dependency relations.