Epistemic Answer Set Programming

Aus International Center for Computational Logic
Wechseln zu:Navigation, Suche

Epistemic Answer Set Programming

Vortrag von Ezgi Iraz Su
Today it is widely accepted by the logic programming community that answer set programming (ASP) requires a more powerful introspective reasoning with the use of modalities. Although there has been a long-lasting debate among researchers about how to correctly extend ASP with epistemic modal operators, there is still no agreement on a fully satisfactory semantics that is able to offer intuitive results for epistemic logic programs. In this talk, we introduce a recent epistemic extension of ASP called epistemic ASP (EASP), endowed with the epistemic answer set semantics: minimal (with respect to truth) models which are maximal under two different orderings that minimise knowledge. Then we compare EASP with existing successful (to some extent) approaches in the literature, showing the advantages and the novelties of the new semantics: compared to Gelfond's epistemic specifications (ES), EASP defines a sufficiently strong language of a simpler syntactic character. Its semantics through a minimality criterion with respect to truth and knowledge is a natural and conservative generalisation of ASP's original answer set semantics. Moreover, compared to all other semantics proposals for ES, the epistemic answer set semantics provides a comprehensive solution to unintended results for epistemic logic programs including constraints. Finally, we briefly discuss some formal properties of EASP such as epistemic splitting, strong equivalence and foundedness.


Speaker info: Ezgi Iraz Su, IRIT (Lilac), Université de Toulouse 3 (Université Paul Sabatier)