Phan Minh Dung published in 1995 in the journal Artificial Intelligence the paper 'On the Acceptability of Arguments and its Fundamental Role in Nonmonotonic Reasoning, Logic Programming and n-Person Games', introducing the argumentation framework (AF): a formal model for representing arguments as nodes and attacks between arguments as edges of a directed graph. The system defines four argumentation semantics (grounded, stable, preferred, complete) to determine which sets of arguments are 'acceptable' in the presence of conflict — without requiring global consistency. Dung's AF provides the mathematical foundation for non-monotonic reasoning in AI: systems that can revise their conclusions in light of new contradictory information without collapsing. It is the formal basis of modern argumentative AI, automated legal reasoning, multi-agent systems, and explainability mechanisms in deep learning. Its connection to paraconsistent logic is documented: Dung's AF operates as an implicit paraconsistent system, allowing locally inconsistent sets of arguments without global trivialization.