WebDevJudge: Evaluating (M)LLMs as Critiques for Web Development Quality

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Li, Chunyang, Zheng, Yilun, Huang, Xinting, Fang, Tianqing, Xu, Jiahao, Chen, Lihui, Song, Yangqiu, Hu, Han
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915829508997120
author Li, Chunyang
Zheng, Yilun
Huang, Xinting
Fang, Tianqing
Xu, Jiahao
Chen, Lihui
Song, Yangqiu
Hu, Han
author_facet Li, Chunyang
Zheng, Yilun
Huang, Xinting
Fang, Tianqing
Xu, Jiahao
Chen, Lihui
Song, Yangqiu
Hu, Han
contents The paradigm of LLM-as-a-judge is emerging as a scalable and efficient alternative to human evaluation, demonstrating strong performance on well-defined tasks. However, its reliability in open-ended tasks with dynamic environments and complex interactions remains unexplored. To bridge the gap, we introduce WebDevJudge, a systematic benchmark for assessing LLM-as-a-judge performance in web development, with support for both non-interactive evaluation based on static observations and continuous interactive evaluation with a dynamic web environment. WebDevJudge comprises human preference labels over paired web implementations, annotated with structured and query-grounded rubrics to ensure high-quality ground truth. Using this benchmark, we comprehensively evaluate various evaluators, including LLMs, MLLMs, and agentic workflows. We systematically investigate the impact of different paradigms and guidance mechanisms. Our experiments reveal a significant gap between LLM judges and human experts. In-depth analysis indicates this gap stems from fundamental model limitations, including failures in recognizing functional equivalence, verifying task feasibility, and mitigating bias. Overall, WebDevJudge presents a challenge to LLM-as-a-judge, offering insights to guide future research toward developing more reliable and capable automated evaluators for complicated scenarios. Code and data are available at https://github.com/lcy2723/WebDevJudge.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18560
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle WebDevJudge: Evaluating (M)LLMs as Critiques for Web Development Quality
Li, Chunyang
Zheng, Yilun
Huang, Xinting
Fang, Tianqing
Xu, Jiahao
Chen, Lihui
Song, Yangqiu
Hu, Han
Software Engineering
Artificial Intelligence
The paradigm of LLM-as-a-judge is emerging as a scalable and efficient alternative to human evaluation, demonstrating strong performance on well-defined tasks. However, its reliability in open-ended tasks with dynamic environments and complex interactions remains unexplored. To bridge the gap, we introduce WebDevJudge, a systematic benchmark for assessing LLM-as-a-judge performance in web development, with support for both non-interactive evaluation based on static observations and continuous interactive evaluation with a dynamic web environment. WebDevJudge comprises human preference labels over paired web implementations, annotated with structured and query-grounded rubrics to ensure high-quality ground truth. Using this benchmark, we comprehensively evaluate various evaluators, including LLMs, MLLMs, and agentic workflows. We systematically investigate the impact of different paradigms and guidance mechanisms. Our experiments reveal a significant gap between LLM judges and human experts. In-depth analysis indicates this gap stems from fundamental model limitations, including failures in recognizing functional equivalence, verifying task feasibility, and mitigating bias. Overall, WebDevJudge presents a challenge to LLM-as-a-judge, offering insights to guide future research toward developing more reliable and capable automated evaluators for complicated scenarios. Code and data are available at https://github.com/lcy2723/WebDevJudge.
title WebDevJudge: Evaluating (M)LLMs as Critiques for Web Development Quality
topic Software Engineering
Artificial Intelligence
url https://arxiv.org/abs/2510.18560