A logic of defeasible argumentation: Constructing arguments in justification logic

In the 1980s, Pollock’s work on default reasons started quest in AI community for a formal system of defeasible argumentation. The main goal this paper is to provide logic structured arguments using language justification logic. logic, we introduce assertions type t : F that read as “t reason justifies F”. Such formulas are then interpreted and their acceptance semantics given analogy Dung’s abstract argumentation framework semantics. We show large subclass frameworks call “warranted” special case our sense (1) can be obtained from logic-based theories by focusing single aspect attacks among (2) warranted always have multiple instantiations called “realizations”. first define new relies operational One key features absent standard logics possibility weigh different epistemic or pieces evidence might conflict with one another. To amend this, develop “defeaters”: conflicting forming basis doubt original conclusion believe an opposite statement. This enables us formalize non-monotonic justifications prompt extension revision already normal theories. Then present arguments. format overlaps idea between extent it possible all notions extensions. Using definitions extensions, establish correspondence results shows notorious attack cycles cannot realized