A New Proposal for Debugging Datalog Programs

In this paper, we propose to apply declarative debugging to Datalog programs. Our approach relies on program semantics rather than on the computation mechanism. The debugging process starts when the user detects an unexpected answer. By asking questions about the intended semantics, the debugger looks for incorrect program relations. While usual declarative debuggers for other languages are based on computation trees, we show that graphs are more convenient structures for representing Datalog computations. The theoretical framework is complemented by the implementation of a debugger for the deductive database system DES, a publicly available open-source project.