ASAP 2.0 dataset¶
Product or Programme · All products and programmes
Evidence · 3 claims (2 for, 1 mixed) · 2 studies (2 associational),q2· 0 of 2 report an effect size · 3 claims rest on one study
Description¶
The ASAP 2.0 dataset is a public benchmark of approximately 24,000 student argumentative essays with human-assigned scores from 1 to 6, developed for automated essay-scoring research.
Components¶
- ASAP 2.0 dataset of argumentative student essays with 1–6 human-assigned scores: The ASAP 2.0 dataset (Crossley et al. 2024) is the public AES benchmark used to evaluate RACES. It "contains approximately 24,000 argumentative essays written by students, covering diverse topics and writing styles", with about 17,000 used for training and the rest for validation and testing. Each essay carries a human-assigned score from 1 to 6 based on strict evaluation criteria, serving as the only human-assigned labels in the experiments. (Zhenxin Zhang et al. (2026))
Claims¶
- Out-of-the-box LLMs agree with human essay grades far less than human raters agree with each other [+W]
- Newer LLM generations do not consistently agree better with human essay graders [+W]
- In Support & Evidence and Claim & Focus, teacher and AES scores agreed at the lowest and highest rubric points but diverged in the middle, with teachers more likely to score a 2 and the AES a 3 [~W]
Related Products and Programmes¶
Key Sources¶
- Zhenxin Zhang, Ziyu Ding, Mengyun Liu and Haiwei Sang. (2026). RACES: reward-aligned consistent essay scoring with large language models. International Journal of Educational Technology in Higher Education. https://doi.org/10.1186/s41239-026-00607-8
Page metadata
| Field | Value |
|---|---|
type |
product |
id |
asap-20-dataset |
title |
ASAP 2.0 dataset |
description |
The ASAP 2.0 dataset is a public benchmark of approximately 24,000 student argumentative essays with human-assigned scores from 1 to 6, developed for automated essay-scoring research. |
status |
review |
product_kind |
dataset |
generated |
by process:settle-candidates on 2026-10-12 |
trust |
unverified — no one has confirmed this page's content |
Frontmatter is the page's machine-readable record — see the Schema & Guide.