One-Way Model Transformations in the Context of the Technology-Roadmapping Tool IRIS

Abstract

CONTEXT: Today’s industry heavily depends on the collaboration of multiple companies that have to exchange a lot of information. When these companies use technology roadmapping and model-driven development, this leads to the exchange of very large models, which is governed by competition laws, and the companies’ interest in protecting their intellectual property. OBJECTIVE: We suggest an automated one-way model transformation approach to remove content worth protecting from a given model in the context of the graphical technology-roadmapping modeling tool IRIS. METHOD: We elaborate constraints that an obfuscated model must fulfill. Based on these, we systematically identify and analyze appropriate one-way transformations to enable the desired level of obfuscation, implementing a proof of concept implementation in IRIS. Our approach first transforms the model into a flattened constraint system and uses nine selected transformations—ranging from constant folding to the tailoring of expressions—to create a new model without the sensitive content. We evaluate our transformations according to a set of predefined properties and constraints regarding their effects on the behavior of the model and the inferable information. LIMITATIONS: Our obfuscation concerns the model and the embedded formulas but does not deal with individual data, which may be subject to privacy-related issues. RESULTS: We find that the identified transformations are applicable in practice and that they can remove a lot of sensitive information from a given model. Future work includes dealing with individual data and protecting against reverse-engineering attacks by domain experts.