In modern modeling tools used for model-driven development, the validation of several well-formedness constraints is continuously been carried out by exploiting advanced graph query engines to highlight conceptual design aws. However, while models are still under development, they are frequently par- tial and incomplete. Validating constraints on incomplete, partial models may identify a large number of irrelevant problems. By switching o the val- idation of these constraints, one may fail to reveal problematic cases which are dicult to correct when the model becomes suciently detailed. Here, we propose a novel validation technique for evaluating well-formed- ness constraints on incomplete, partial models with may and must semantics, e.g. a constraint without a valid match is satisable if there is a completion of the partial model that may satisfy it. To this end, we map the problem of constraint evaluation over partial models into regular graph pattern matching over complete models by semantically equivalent rewrites of graph queries.
Robust Confidence Regions for Incomplete Models