Single-Language Evidence Is Insufficient for Automated Logging: A Multilingual Benchmark and Empirical Study with LLMs DOI: XXXXXXX.XXXXXXX CCS: Software and its engineering Maintaining Software

Renyi Zhong, Yichen Li, Yulun Wu, Jinxi Kuang, Yintong Huo, Michael R. Lyu · 2026

Read Single-Language Evidence Is Insufficient for Automated Logging: A Multilingual Benchmark and Empirical Study with LLMs DOI: XXXXXXX.XXXXXXX CCS: Software and its engineering Maintaining Software by Renyi Zhong, Yichen Li, Yulun Wu, Jinxi Kuang, Yintong Huo, Michael R. Lyu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-17529.txt.

Description

Zhong and five colleagues argue that automated logging research leans too hard on Java, leaving open whether its lessons travel to other languages. Their MultiLogBench spans six language ecosystems with 63,965 production-code snapshots, 744 revision-history cases where developers added logging during maintenance, and a paired transformed branch for robustness checks. Testing seven current large language models under one protocol, they cover logging-site location, framework matching, severity prediction, and message writing. The study maps where model behavior and model choice hold up across languages and code change.

Catalog record

CREATORRenyi Zhong, Yichen Li, Yulun Wu, Jinxi Kuang, Yintong Huo, Michael R. Lyu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE106255
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.17529

Browse the full library · About the Tower of Babel