Summary of Entity-level Factual Adaptiveness Of Fine-tuning Based Abstractive Summarization Models, by Jongyoon Song et al.
Entity-level Factual Adaptiveness of Fine-tuning based Abstractive Summarization Models
by Jongyoon Song, Nohil Park, Bongkyu Hwang, Jaewoong Yun, Seongho Joe, Youngjune L. Gwon, Sungroh Yoon
First submitted to arxiv on: 23 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Machine learning researchers have long struggled to create summarization models that generate accurate and consistent summaries. A major issue arises when a model’s knowledge conflicts with the knowledge in the input document, leading to factually inconsistent content. In this paper, the authors analyze the robustness of fine-tuning based summarization models to these knowledge conflicts, which they call factual adaptiveness. They find that while some models perform well on original datasets, their ability to adapt to new information is limited. To address this issue, the researchers introduce a controllable data augmentation method that allows for adjustable levels of knowledge conflict. This approach enhances the model’s ability to generate accurate and consistent summaries. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Summarization models often struggle to create accurate and consistent summaries because they may have conflicting knowledge with the information in the input document. Researchers are trying to find ways to make these models better at handling this kind of information. In this study, scientists tested how well some summarization models can handle different types of knowledge conflicts. They found that some models do better than others on original datasets but struggle when dealing with new information. To help models work better in the future, researchers developed a way to adjust the amount of conflicting information they are given. |
Keywords
* Artificial intelligence * Data augmentation * Fine tuning * Machine learning * Summarization
