@inproceedings{xu-yan-2025-triplefact,
title = "{T}riple{F}act: Defending Data Contamination in the Evaluation of {LLM}-driven Fake News Detection",
author = "Xu, Cheng and
Yan, Nan",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.acl-long.431/",
doi = "10.18653/v1/2025.acl-long.431",
pages = "8808--8823",
ISBN = "979-8-89176-251-0",
abstract = "The proliferation of large language models (LLMs) has introduced unprecedented challenges in fake news detection due to benchmark data contamination (BDC), where evaluation benchmarks are inadvertently memorized during the pre-training, leading to the inflated performance metrics. Traditional evaluation paradigms, reliant on static datasets and closed-world assumptions, fail to account the BDC risk in large-scale pre-training of current LLMs. This paper introduces TripleFact, a novel evaluation framework for fake news detection task, which designed to mitigate BDC risk while prioritizing real-world applicability. TripleFact integrates three components: (1) Human-Adversarial Preference Testing (HAPT) to assess robustness against human-crafted misinformation, (2) Real-Time Web Agent with Asynchronous Validation (RTW-AV) to evaluate temporal generalization using dynamically sourced claims, and (3) Entity-Controlled Virtual Environment (ECVE) to eliminate entity-specific biases. Through experiments on 17 state-of-the-art LLMs, including GPT, LLaMA, and DeepSeek variants, TripleFact demonstrates superior contamination resistance compared to traditional benchmarks. Results reveal that BDC artificially inflates performance by up to 23{\%} in conventional evaluations, while TripleFact Score (TFS) remain stable within 4{\%} absolute error under controlled contamination. The framework{'}s ability to disentangle genuine detection capabilities from memorization artifacts underscores its potential as a fake news detection benchmark for the LLM era."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="xu-yan-2025-triplefact">
<titleInfo>
<title>TripleFact: Defending Data Contamination in the Evaluation of LLM-driven Fake News Detection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Cheng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nan</namePart>
<namePart type="family">Yan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-251-0</identifier>
</relatedItem>
<abstract>The proliferation of large language models (LLMs) has introduced unprecedented challenges in fake news detection due to benchmark data contamination (BDC), where evaluation benchmarks are inadvertently memorized during the pre-training, leading to the inflated performance metrics. Traditional evaluation paradigms, reliant on static datasets and closed-world assumptions, fail to account the BDC risk in large-scale pre-training of current LLMs. This paper introduces TripleFact, a novel evaluation framework for fake news detection task, which designed to mitigate BDC risk while prioritizing real-world applicability. TripleFact integrates three components: (1) Human-Adversarial Preference Testing (HAPT) to assess robustness against human-crafted misinformation, (2) Real-Time Web Agent with Asynchronous Validation (RTW-AV) to evaluate temporal generalization using dynamically sourced claims, and (3) Entity-Controlled Virtual Environment (ECVE) to eliminate entity-specific biases. Through experiments on 17 state-of-the-art LLMs, including GPT, LLaMA, and DeepSeek variants, TripleFact demonstrates superior contamination resistance compared to traditional benchmarks. Results reveal that BDC artificially inflates performance by up to 23% in conventional evaluations, while TripleFact Score (TFS) remain stable within 4% absolute error under controlled contamination. The framework’s ability to disentangle genuine detection capabilities from memorization artifacts underscores its potential as a fake news detection benchmark for the LLM era.</abstract>
<identifier type="citekey">xu-yan-2025-triplefact</identifier>
<identifier type="doi">10.18653/v1/2025.acl-long.431</identifier>
<location>
<url>https://aclanthology.org/2025.acl-long.431/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>8808</start>
<end>8823</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TripleFact: Defending Data Contamination in the Evaluation of LLM-driven Fake News Detection
%A Xu, Cheng
%A Yan, Nan
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-251-0
%F xu-yan-2025-triplefact
%X The proliferation of large language models (LLMs) has introduced unprecedented challenges in fake news detection due to benchmark data contamination (BDC), where evaluation benchmarks are inadvertently memorized during the pre-training, leading to the inflated performance metrics. Traditional evaluation paradigms, reliant on static datasets and closed-world assumptions, fail to account the BDC risk in large-scale pre-training of current LLMs. This paper introduces TripleFact, a novel evaluation framework for fake news detection task, which designed to mitigate BDC risk while prioritizing real-world applicability. TripleFact integrates three components: (1) Human-Adversarial Preference Testing (HAPT) to assess robustness against human-crafted misinformation, (2) Real-Time Web Agent with Asynchronous Validation (RTW-AV) to evaluate temporal generalization using dynamically sourced claims, and (3) Entity-Controlled Virtual Environment (ECVE) to eliminate entity-specific biases. Through experiments on 17 state-of-the-art LLMs, including GPT, LLaMA, and DeepSeek variants, TripleFact demonstrates superior contamination resistance compared to traditional benchmarks. Results reveal that BDC artificially inflates performance by up to 23% in conventional evaluations, while TripleFact Score (TFS) remain stable within 4% absolute error under controlled contamination. The framework’s ability to disentangle genuine detection capabilities from memorization artifacts underscores its potential as a fake news detection benchmark for the LLM era.
%R 10.18653/v1/2025.acl-long.431
%U https://aclanthology.org/2025.acl-long.431/
%U https://doi.org/10.18653/v1/2025.acl-long.431
%P 8808-8823
Markdown (Informal)
[TripleFact: Defending Data Contamination in the Evaluation of LLM-driven Fake News Detection](https://aclanthology.org/2025.acl-long.431/) (Xu & Yan, ACL 2025)
ACL