Summary of Autoregressive Score Generation For Multi-trait Essay Scoring, by Heejin Do et al.
Autoregressive Score Generation for Multi-trait Essay Scoring
by Heejin Do, Yunsu Kim, Gary Geunbae Lee
First submitted to arxiv on: 13 Mar 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 Recently, pre-trained models like BERT have been applied to automated essay scoring (AES) with great success, predicting a single overall score. However, these models have yet to be explored for multi-trait AES, likely due to the inefficiency of replicating them for each trait. This paper proposes an autoregressive prediction model called ArTS, which incorporates a decoding process using pre-trained T5. Unlike previous methods, this approach redefines AES as a score-generation task, allowing a single model to predict multiple scores. During decoding, subsequent trait predictions benefit from conditioning on preceding scores. Experimental results demonstrate the efficacy of ArTS, showing average improvements of over 5% in both prompts and traits. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about using artificial intelligence (AI) to grade essays that have multiple good or bad points. Right now, AI models like BERT are really good at giving one overall score for an essay. But this paper shows how we can make these models better by letting them predict multiple scores for different things in the essay. This new approach is called ArTS and it uses a model called T5 to make predictions. The results show that this method works really well, improving grades by over 5%. |
Keywords
» Artificial intelligence » Autoregressive » Bert » T5
