RST-LoRA: A Discourse-Aware Low-Rank Adaptation for Long Document Abstractive Summarization

arXiv:2405.00657v1 Announce Type: cross
Abstract: For long document summarization, discourse structure is important to discern the key content of the text and the differences in importance level between sentences. Unfortunately, the integration of rhetorical structure theory (RST) into parameter-efficient fine-tuning strategies for long document summarization remains unexplored. Therefore, this paper introduces RST-LoRA and proposes four RST-aware variants to explicitly incorporate RST into the LoRA model. Our empirical evaluation demonstrates that incorporating the type and uncertainty of rhetorical relations can complementarily …

abstract arxiv cs.ai cs.cl cs.lg differences discourse document fine-tuning importance integration key lora low low-rank adaptation paper strategies summarization text the key theory type