Summary of Evaluating Evidence Attribution in Generated Fact Checking Explanations, by Rui Xing et al.
Evaluating Evidence Attribution in Generated Fact Checking Explanations
by Rui Xing, Timothy Baldwin, Jey Han Lau
First submitted to arxiv on: 18 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 This research paper proposes a novel approach to fact-checking explanation generation, addressing the issue of trustworthiness in automated systems. The authors introduce an evaluation protocol called citation masking and recovery, which assesses the quality of attribution in generated explanations using both human annotators and automatic annotators. The results show that while LLMs can be used for annotation, they still generate explanations with inaccurate attributions, emphasizing the importance of human-curated evidence for improving explanation quality. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Automated fact-checking systems are great, but sometimes they get it wrong! This paper helps fix this problem by figuring out how to show where facts come from in their explanations. They came up with a special way to test if explanations are accurate or not. Surprisingly, even the best computer models still make mistakes! To get better results, humans need to be involved. |
