RDF data transformations are transformations of RDF graphs to RDF graphs which preserve in different degree the data content in the source to the target. These transformation therefore give special attention to the data elements in such graphs—under the assumption that data elements reside in the subjects and objects of RDF triples, and the peculiar fact that the set of vertices and set of edges in an RDF graph are not necessarily disjoint. Bounded homomorphisms are used to define these transformations, which not only ensure that data from the source is structurally preserved in the target, but also require, in various ways, the target data to be related back to the source. The result of this paper is a theoretical toolkit of transformation characteristics with which detailed control over the transformation target may be exercised. We explore these characteristics in two different RDF graph representations, and give an algorithm for checking existence of transformations.