A framework for accomplishing vocabulary-based data transformations, for interfacing systems of disparate data formats. The framework is layered to separate the terms of documents form the underlying concepts corresponding thereto. The layers include an ontology layer for defining concepts, a vocabulary layer for labeling the concepts, a specification layer that specifies the concepts and labels, and a representation layer for encoding data in conformance with the specification layer. The separation enables the relationships between existing transformations to be automatically leveraged for minimizing user input for creating new transformations.