Modeling Variations of First-Order Horn Abduction in Answer Set Programming

What is it about?

We formalize a framework for interpretation of natural language by abduction. We realize this framework in the logic programming framework of Answer Set Programming. Our realization has several variations that have different levels of expressivity (cyclic versus acyclic theories). We also show empirical evaluation of our encodings and present on-demand instantiation of certain constraints as a method to significantly increase performance of our method.

Featured Image

Why is it important?

The framework by Hobbs et al. has not been formally described in a declarative way as done in this work. The ASP encodings provide fine-grained control over the depth of reasoning. Finally efficiency is provided using lazy instantiation of constraints which is a quite novel topic in ASP methodology.