Computational Argumentation

The field models arguments, evidence, and conflicts—often as graphs of attacks and support—and uses semantics such as grounded, preferred, and gradual methods to determine which claims hold or how strongly they are supported. It also develops tools for argument mining, stance and relation detection, debate analysis, and applications including law, policy, databases, and decision support. Current approaches combine symbolic frameworks and solvers with neural models and large language models that extract, generate, or evaluate arguments, using benchmarks ranging from annotated essays and debates to formal reasoning tasks. A central challenge is making systems scale to implicit, context-dependent natural language while retaining the reliability, transparency, and guarantees of formal reasoning.