Summary of Reference-based Metrics Disprove Themselves in Question Generation, by Bang Nguyen et al.
Reference-based Metrics Disprove Themselves in Question Generation
by Bang Nguyen, Mengxia Yu, Yun Huang, Meng Jiang
First submitted to arxiv on: 18 Mar 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 The paper presents a study on evaluating question generation (QG) models using reference-based metrics like BLEU and BERTScore. The authors find that these metrics are not reliable when used on QG benchmarks such as SQuAD and HotpotQA, where a single human-written reference is often provided. To address this limitation, they propose a new reference-free metric that considers multi-dimensional criteria like naturalness, answerability, and complexity, leveraging large language models. The authors demonstrate the effectiveness of their proposed metric in accurately distinguishing between high-quality questions and flawed ones. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The study explores how to evaluate question generation (QG) models. Researchers use two types of metrics: those based on human-written references and a new approach that doesn’t rely on these references. They tested both methods on popular QG benchmarks and found that the reference-based metrics didn’t always work well. To fix this, they developed a new metric that looks at different qualities of questions, like how natural or answerable they are. This new method is better than the old one and helps identify good and bad questions more accurately. |
Keywords
* Artificial intelligence * Bleu
