Rhetorical Structure

Rhetorical structure analysis focuses on computationally understanding and modeling the persuasive and stylistic elements within text, aiming to improve natural language processing tasks and offer insights into human communication. Current research emphasizes developing datasets for various rhetorical devices (e.g., parallelism, metaphor, appeals to ethos and pathos), employing large language models (LLMs) and transformer-based architectures for detection and generation, and exploring hierarchical models to capture complex argumentative structures. This work has implications for diverse applications, including automated writing evaluation, argument mining, bias detection in news, and even educational tools to improve teaching practices by identifying and promoting growth mindset language.