On Meta-Evaluation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Hongxiao, Wang, Chenxi, Fan, Fanda, Wang, Zihan, Gao, Wanling, Wang, Lei, Zhan, Jianfeng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914268406874112
author Li, Hongxiao
Wang, Chenxi
Fan, Fanda
Wang, Zihan
Gao, Wanling
Wang, Lei
Zhan, Jianfeng
author_facet Li, Hongxiao
Wang, Chenxi
Fan, Fanda
Wang, Zihan
Gao, Wanling
Wang, Lei
Zhan, Jianfeng
contents Evaluation is the foundation of empirical science, yet the evaluation of evaluation itself -- so-called meta-evaluation -- remains strikingly underdeveloped. While methods such as observational studies, design of experiments (DoE), and randomized controlled trials (RCTs) have shaped modern scientific practice, there has been little systematic inquiry into their comparative validity and utility across domains. Here we introduce a formal framework for meta-evaluation by defining the evaluation space, its structured representation, and a benchmark we call AxiaBench. AxiaBench enables the first large-scale, quantitative comparison of ten widely used evaluation methods across eight representative application domains. Our analysis reveals a fundamental limitation: no existing method simultaneously achieves accuracy and efficiency across diverse scenarios, with DoE and observational designs in particular showing significant deviations from real-world ground truth. We further evaluate a unified method of entire-space stratified sampling from previous evaluatology research, and the results report that it consistently outperforms prior approaches across all tested domains. These results establish meta-evaluation as a scientific object in its own right and provide both a conceptual foundation and a pragmatic tool set for advancing trustworthy evaluation in computational and experimental research.
format Preprint
id arxiv_https___arxiv_org_abs_2601_14262
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle On Meta-Evaluation
Li, Hongxiao
Wang, Chenxi
Fan, Fanda
Wang, Zihan
Gao, Wanling
Wang, Lei
Zhan, Jianfeng
Methodology
Artificial Intelligence
Human-Computer Interaction
Machine Learning
Evaluation is the foundation of empirical science, yet the evaluation of evaluation itself -- so-called meta-evaluation -- remains strikingly underdeveloped. While methods such as observational studies, design of experiments (DoE), and randomized controlled trials (RCTs) have shaped modern scientific practice, there has been little systematic inquiry into their comparative validity and utility across domains. Here we introduce a formal framework for meta-evaluation by defining the evaluation space, its structured representation, and a benchmark we call AxiaBench. AxiaBench enables the first large-scale, quantitative comparison of ten widely used evaluation methods across eight representative application domains. Our analysis reveals a fundamental limitation: no existing method simultaneously achieves accuracy and efficiency across diverse scenarios, with DoE and observational designs in particular showing significant deviations from real-world ground truth. We further evaluate a unified method of entire-space stratified sampling from previous evaluatology research, and the results report that it consistently outperforms prior approaches across all tested domains. These results establish meta-evaluation as a scientific object in its own right and provide both a conceptual foundation and a pragmatic tool set for advancing trustworthy evaluation in computational and experimental research.
title On Meta-Evaluation
topic Methodology
Artificial Intelligence
Human-Computer Interaction
Machine Learning
url https://arxiv.org/abs/2601.14262