Discourse Parser

Discourse parsing aims to automatically identify the structural relationships between sentences in a text, revealing its underlying discourse organization. Current research focuses on improving the accuracy and efficiency of discourse parsers, often leveraging large language models and exploring various architectures like dependency frameworks and gradient boosting, while addressing data scarcity through techniques such as distant supervision and cross-lingual training. These advancements are crucial for enhancing numerous NLP applications, including improved text summarization, question answering, and machine translation, by providing richer contextual understanding of text. The development of larger, more diverse, and multi-genre discourse treebanks is a key area of ongoing effort.