Identifying patch correctness in test-based program repair

Abstract

Test-based automatic program repair has attracted a lot of attention in recent years. However, the test suites in practice are often too weak to guarantee correctness and existing approaches often generate a large number of incorrect patches.