Abstract
In this presentation, I will provide a practical overview of the Rational Speech Act framework, a probabilistic model of pragmatic language understanding. The RSA framework captures how speakers and listeners recursively reason about each other’s interpretations of utterances, starting from literal semantics and incorporating goals, costs, and contextual priors. Through examples such as simple objects and vague adjectives, I will demonstrate how RSA models explain phenomena like hyperbole, politeness, and lexical uncertainty. Extensions to the framework address compositionality, adaptation, and epistemic inference, offering insights into both human cognition and language model systems.
Duc Q. Nguyen
CS PhD Student
My research interests include Generative Models, Graph Representation Learning, and Probabilistic Machine Learning. My application interests include Natural Language Processing, Healthcare, and Education.