@inproceedings{lin-etal-2026-melonbench,
title = "{M}elon{B}ench: A Benchmark for Coded Entity Resolution over Noisy Web Evidence",
author = "Lin, Zhishan and
Li, Jingxiao and
Zhang, Yanbing and
Liu, Ziyan and
Shao, Yuqing and
Tong, Yunfei and
Liu, Ganghao and
Zhang, Qian and
Wang, Zhe",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.660/",
pages = "15202--15233",
ISBN = "979-8-89176-459-0",
abstract = "Large language models (LLMs) are increasingly deployed as web-browsing agents, yet current benchmarks mostly evaluate explicit queries over clean factual evidence. We introduce Melonbench, a 665-item benchmark for coded entity resolution over noisy open-web user-generated content, instantiated with mid-to-long-tail public figures in Chinese entertainment blind-item discourse. Solving Melonbench requires agents to decode obfuscated clues, retrieve fragmented evidence, and identify the target entity. The benchmark is constructed through a five-stage semi-automatic pipeline with atomic event stratification, adversarial faithfulness auditing, and expert verification. Across 30 systems, the best non-search model reaches 25.1{\%} accuracy and the best search-enabled system reaches 49.9{\%}. Controlled comparisons of the same models with and without web search confirm consistent gains from web search, but also substantial remaining headroom. Diagnostic experiments identify semantic de-obfuscation and open-ended entity recall as key bottlenecks, and show that reasoning-centric search can outperform report-oriented agents. An 80-item finance pilot provides preliminary evidence of transfer beyond entertainment within a Chinese-language setting. Melonbench offers a challenging initial testbed for search agents under intentional obfuscation and noisy evidence. The dataset is publicly available at https://huggingface.co/datasets/janehaha/MelonBench."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lin-etal-2026-melonbench">
<titleInfo>
<title>MelonBench: A Benchmark for Coded Entity Resolution over Noisy Web Evidence</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhishan</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingxiao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanbing</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ziyan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuqing</namePart>
<namePart type="family">Shao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunfei</namePart>
<namePart type="family">Tong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ganghao</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qian</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhe</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Large language models (LLMs) are increasingly deployed as web-browsing agents, yet current benchmarks mostly evaluate explicit queries over clean factual evidence. We introduce Melonbench, a 665-item benchmark for coded entity resolution over noisy open-web user-generated content, instantiated with mid-to-long-tail public figures in Chinese entertainment blind-item discourse. Solving Melonbench requires agents to decode obfuscated clues, retrieve fragmented evidence, and identify the target entity. The benchmark is constructed through a five-stage semi-automatic pipeline with atomic event stratification, adversarial faithfulness auditing, and expert verification. Across 30 systems, the best non-search model reaches 25.1% accuracy and the best search-enabled system reaches 49.9%. Controlled comparisons of the same models with and without web search confirm consistent gains from web search, but also substantial remaining headroom. Diagnostic experiments identify semantic de-obfuscation and open-ended entity recall as key bottlenecks, and show that reasoning-centric search can outperform report-oriented agents. An 80-item finance pilot provides preliminary evidence of transfer beyond entertainment within a Chinese-language setting. Melonbench offers a challenging initial testbed for search agents under intentional obfuscation and noisy evidence. The dataset is publicly available at https://huggingface.co/datasets/janehaha/MelonBench.</abstract>
<identifier type="citekey">lin-etal-2026-melonbench</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.660/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>15202</start>
<end>15233</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MelonBench: A Benchmark for Coded Entity Resolution over Noisy Web Evidence
%A Lin, Zhishan
%A Li, Jingxiao
%A Zhang, Yanbing
%A Liu, Ziyan
%A Shao, Yuqing
%A Tong, Yunfei
%A Liu, Ganghao
%A Zhang, Qian
%A Wang, Zhe
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F lin-etal-2026-melonbench
%X Large language models (LLMs) are increasingly deployed as web-browsing agents, yet current benchmarks mostly evaluate explicit queries over clean factual evidence. We introduce Melonbench, a 665-item benchmark for coded entity resolution over noisy open-web user-generated content, instantiated with mid-to-long-tail public figures in Chinese entertainment blind-item discourse. Solving Melonbench requires agents to decode obfuscated clues, retrieve fragmented evidence, and identify the target entity. The benchmark is constructed through a five-stage semi-automatic pipeline with atomic event stratification, adversarial faithfulness auditing, and expert verification. Across 30 systems, the best non-search model reaches 25.1% accuracy and the best search-enabled system reaches 49.9%. Controlled comparisons of the same models with and without web search confirm consistent gains from web search, but also substantial remaining headroom. Diagnostic experiments identify semantic de-obfuscation and open-ended entity recall as key bottlenecks, and show that reasoning-centric search can outperform report-oriented agents. An 80-item finance pilot provides preliminary evidence of transfer beyond entertainment within a Chinese-language setting. Melonbench offers a challenging initial testbed for search agents under intentional obfuscation and noisy evidence. The dataset is publicly available at https://huggingface.co/datasets/janehaha/MelonBench.
%U https://aclanthology.org/2026.emnlp-main.660/
%P 15202-15233
Markdown (Informal)
[MelonBench: A Benchmark for Coded Entity Resolution over Noisy Web Evidence](https://aclanthology.org/2026.emnlp-main.660/) (Lin et al., EMNLP 2026)
ACL
- Zhishan Lin, Jingxiao Li, Yanbing Zhang, Ziyan Liu, Yuqing Shao, Yunfei Tong, Ganghao Liu, Qian Zhang, and Zhe Wang. 2026. MelonBench: A Benchmark for Coded Entity Resolution over Noisy Web Evidence. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 15202–15233, Budapest, Hungary. Association for Computational Linguistics.