A Versatile Intermediate Language for Answer Set Programming

The attractiveness of Answer Set Programming (ASP) and related paradigms for declarative problem solving is considerably due to the availability of highly efficient yet easy-to-use implementations. A major driving force for the development and improvement of tools are standardized problem representations, for several reasons. First, they relieve developers from the burden of inventing their own input formats. Second, they establish interoperability between separate tools, allowing users to easily compare and exchange them without extensively converting their problem representations. Third, they facilitate the acquisition of problem descriptions from distinct sources, which is useful for benchmarking and assessment purposes. Historically, however, standards for representing logic programs, serving as inputs to ASP systems, were mainly dictated by the few available tools. In fact, there currently are two quasi standards, namely, the formats used by lparse and dlv, incompatible with each other. As a first step towards overcoming this deficiency, this work proposes an intermediate format for ground logic programs, intended for the representation of inputs to ASP solvers. The format is not designed to be a primary input language, given that ASP systems usually deploy a second component, called a grounder, to deal with the inputs provided by users. In view of this, our format is situated intermediate a grounder and a solver, guided by the example of grounder lparse and solver smodels, the latter marking the first among nowadays a variety of solvers processing the output of lparse. However, the output format of lparse has some decisive drawbacks, namely, its restrictive range and limited extensibility. We thus propose a new intermediate language, where our major design goals are flexibility in problem representation and easy extensibility to new language constructs.