SCL Clause Learning from Simple Models
Several decision procedures for the Bernays-Schoenfinkel (BS) fragment of first-order logic rely on explicit model assumptions. In particular, the procedures differ in their respective model representation formalisms. We introduce a new decision procedure SCL deciding the BS fragment. SCL stands for clause learning from simple models. Simple models are solely built on ground literals. Nevertheless, we show that SCL can learn exactly the clauses other procedures learn with respect to more complex model representation formalisms. Therefore, the overhead of complex model representation formalisms is not always needed. SCL is sound and complete for full first-order logic without equality.
This work was funded by DFG grant 389792660 as part of TRR 248.