Question Under Discussion

Questions Under Discussion (QUD) analysis frames discourse as a continuous process of asking and answering questions, aiming to automatically identify the implicit questions driving sentence relationships within a text. Current research focuses on developing robust QUD parsers, often employing joint-training frameworks and selective decoding strategies to improve the accuracy of question generation and anchor sentence identification, addressing challenges in satisfying theoretical QUD criteria. This approach offers a potentially more scalable and accessible alternative to traditional discourse analysis methods, with applications in tasks such as document summarization and simplification.