Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2407.15021 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866916331452891136 |
|---|---|
| author | Hwang, EunJeong Zhou, Yichao Wendt, James Bradley Gunel, Beliz Vo, Nguyen Xie, Jing Tata, Sandeep |
| author_facet | Hwang, EunJeong Zhou, Yichao Wendt, James Bradley Gunel, Beliz Vo, Nguyen Xie, Jing Tata, Sandeep |
| contents | Large language models (LLMs) often struggle with processing extensive input contexts, which can lead to redundant, inaccurate, or incoherent summaries. Recent methods have used unstructured memory to incrementally process these contexts, but they still suffer from information overload due to the volume of unstructured data handled. In our study, we introduce structured knowledge representations ($GU_{json}$), which significantly improve summarization performance by 40% and 14% across two public datasets. Most notably, we propose the Chain-of-Key strategy ($CoK_{json}$) that dynamically updates or augments these representations with new information, rather than recreating the structured memory for each new source. This method further enhances performance by 7% and 4% on the datasets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_15021 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Enhancing Incremental Summarization with Structured Representations Hwang, EunJeong Zhou, Yichao Wendt, James Bradley Gunel, Beliz Vo, Nguyen Xie, Jing Tata, Sandeep Computation and Language Large language models (LLMs) often struggle with processing extensive input contexts, which can lead to redundant, inaccurate, or incoherent summaries. Recent methods have used unstructured memory to incrementally process these contexts, but they still suffer from information overload due to the volume of unstructured data handled. In our study, we introduce structured knowledge representations ($GU_{json}$), which significantly improve summarization performance by 40% and 14% across two public datasets. Most notably, we propose the Chain-of-Key strategy ($CoK_{json}$) that dynamically updates or augments these representations with new information, rather than recreating the structured memory for each new source. This method further enhances performance by 7% and 4% on the datasets. |
| title | Enhancing Incremental Summarization with Structured Representations |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2407.15021 |