Evaluation: later research

design · draft · NLP and argumentation · edited 2026-11-04

Not designed. The current focus is argument structure.

Typing says what kind of argument something is. Evaluation says whether it’s good, and who wins the debate. These are the leads to research later:

• CAPNA (Hinton and Wagemans): judging an argument in stages. capna-based-flow.md sketches response categories from it.

• Dung’s abstract argumentation: arguments as a graph of attacks; “winning” as which arguments survive.

  • Dung-O-Matic and Dungine, Java implementations from ARG-tech: https://arg.tech/research/dung-o-matic/
• Value-based argumentation (Bench-Capon 2003): each NPC ranks values, so the same argument can win against one NPC and fail against another. Ada will get the paper.

What links here: Quell integration, Open questions, Quell debate: plan map, NPC stances: data model, Debate, CAPNA, Abstract argumentation, Value-based argumentation