Hierarchical Indexing for Retrieval-Augmented Opinion Summarization

arXiv:2403.00435v1 Announce Type: new
Abstract: We propose a method for unsupervised abstractive opinion summarization, that combines the attributability and scalability of extractive approaches with the coherence and fluency of Large Language Models (LLMs). Our method, HIRO, learns an index structure that maps sentences to a path through a semantically organized discrete hierarchy. At inference time, we populate the index and use it to identify and retrieve clusters of sentences containing popular opinions from input reviews. Then, we use a pretrained …

abstract arxiv cs.cl hierarchical index indexing inference language language models large language large language models llms maps opinion path retrieval retrieval-augmented scalability summarization through type unsupervised