Skip to content

Rubric-aligned concept-annotated grading benchmarks (REC-CBM)

Product or Programme · All products and programmes
Evidence · no claims cited

Description

REC-CBM is a research dataset release that adds rubric-aligned, expert-validated concept annotations to the Mohler, ASAP 2.0, and MOCHA open-ended response-grading benchmarks for interpretable automated grading research.

Components

  • Rubric-aligned concept-annotated grading benchmarks (Mohler, ASAP 2.0, MOCHA): The authors curate and annotate three public open-ended grading benchmarks with rubric-aligned concept labels: Mohler short-answer grading (2,273 samples, 8 concepts, 3 concept levels, 6 grade levels), ASAP 2.0 essay scoring (17,292 samples, 8 concepts, 5 concept levels, 6 grade levels), and MOCHA reading-comprehension answer grading (31,069 samples, 7 concepts, 3 concept levels, 5 grade levels). None of the three benchmarks originally included concept-level annotations, so the authors create them using a human-in-the-loop pipeline. Code and data are released at https://github.com/scott-f-zhang/REC-CBM and https://huggingface.co/datasets/scott-f-zhang/REC-CBM. (Chengshuai Zhao et al. (2025))

Claims

Key Sources

  • Chengshuai Zhao, Fan Zhang, Kumar Satvik Chaudhary, Yiwen Li, Lo Pang-Yun Ting, Ying-Chih Chen, Huan Liu. (2025). REC-CBM: Rubric-Aware Error-Correction Concept Bottleneck Models for Trustworthy Open-Ended Grading. https://github.com/scott-f-zhang/REC-CBM
Page metadata
Field Value
type product
id rubric-aligned-concept-annotated-grading-benchmarks-rec-cbm
title Rubric-aligned concept-annotated grading benchmarks (REC-CBM)
description REC-CBM is a research dataset release that adds rubric-aligned, expert-validated concept annotations to the Mohler, ASAP 2.0, and MOCHA open-ended response-grading benchmarks for interpretable automated grading research.
status review
product_kind dataset
generated by process:settle-candidates on 2026-10-11
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.