Computing Diverse Optimal Stable Models

We introduce a comprehensive framework for computing diverse (or similar) solutions to logic programs with preferences. Our framework provides a wide spectrum of complete and incomplete methods for solving this task. Apart from proposing several new methods, it also accommodates existing ones and generalizes them to programs with preferences. Interestingly, this is accomplished by integrating and automating several basic ASP techniques - being of general interest even beyond diversification. The enabling factor of this lies in the recent advance of multi-shot ASP solving that provides us with fine-grained control over reasoning processes and abolishes the need for solver modifications and wrappers that were indispensable in previous approaches. Our framework is implemented as an extension to the ASP-based preference handling system asprin. We use the resulting system asprin 2 for an empirical evaluation of the diversification methods comprised in our framework.