PaperlayerView exact revision on arXiv

Source-linked AI summary

How Robust Are Automated Fact-Checking Systems? A Cross-Benchmark Evaluation

Aida Usmanova, Zangir Iklassov, Markus Leippold, Ricardo Usbeck

arXiv:2608.25934v1cs.AIcs.LG

TL;DR

Automated fact-checking evaluations often omit simple baselines and test systems on only one benchmark, leaving cross-domain generalisation and full-pipeline robustness uncertain. This paper evaluates nine models across four diverse datasets and finds domain-dependent rankings, competitive classical baselines, and retrieval as the primary bottleneck. It also documents scope limits involving dataset coverage, model coverage, and prompt sensitivity.

  • Problem

    The paper addresses the lack of cross-domain evaluation of complete retrieve-then-verify AFC pipelines with simple baselines.

    Source: §1 · pp. 1–2, G Recommendations for AFC Evaluation… · p. 21

  • Method

    It benchmarks nine models across four datasets, covering random and sparse baselines, fine-tuned transformers, zero-shot LLMs, and top AVeriTeC 2025 systems.

    Source: §1 · p. 2, §3.3 · p. 3

  • Results

    System rankings depend strongly on domain and metric, classical baselines can outperform evidence-conditioned systems, and retrieval remains the primary bottleneck.

    Source: §1 · p. 2, §4.4 · p. 8, §5 · p. 9

  • Takeaways & Limitations

    The findings support evaluating AFC systems across multiple domains and metrics rather than treating single-benchmark rank as general capability.

    Source: §1 · p. 2, G Recommendations for AFC Evaluation… · p. 21

    Conclusions are limited to four datasets and do not cover multilingual or multi-domain AFC, recent generative systems beyond Llama 3.1, or prompt sensitivity.

    Source: Limitations · p. 9

Abstract

from arXiv · show

Automated fact-checking (AFC) systems retrieve evidence and predict claim veracity, yet evaluations omit simple baselines, systems are developed for a single benchmark and cannot be trusted to generalise across domains. No prior work cross-evaluates the full two-stage retrieve-then-verify pipeline across diverse datasets, complementing retrieval-only studies (Thakur et al., 2021) and single-stage benchmarking studies (Calamai et al., 2025). We benchmark nine models, ranging from random and sparse baselines to fine-tuned transformers, zero-shot LLMs, and the two highest-ranked systems from the AVeriTeC 2025 shared task, across four datasets spanning scientific, open-web, and climate domains. Three findings stand out: (1) on ClimateCheck claim-only and fine-tuned models outperform zero-shot LLM and top-performing AVeriTeC 2025 systems, highlighting that noisy evidence can degrade veracity prediction; (2) system rankings are strongly domain- and metric-dependent: the best model on SciFact (macro-F1 0.70) drops to 0.31 on ClimateCheck, while the AVeriTeC 2025 winner and runner-up swap rankings based on evaluation metrics and datasets; (3) replacing retrieved evidence with gold annotations improves veracity accuracy by 14-22 points across models, confirming retrieval remains primary bottleneck. We release code, pre-processed datasets, and all results to support reproducible AFC research.

1 Introduction

The paper asks whether automated fact-checking progress generalises beyond individual benchmarks and evaluates the complete retrieve-then-verify pipeline across diverse datasets. Its results show strong baseline competitiveness, domain- and metric-dependent rankings, and retrieval as the main bottleneck.

Motivation

  • AFC systems may not generalise across domains because they are usually developed and tested on single benchmarks without simple lower-bound baselines.The paper argues that absent baselines obscure whether gains reflect language understanding or dataset-specific engineering.Source: §1 · p. 1

Research gap

  • Prior studies examined retrieval or single-task classification separately, but no prior work cross-evaluated the complete retrieve-then-verify pipeline across structurally diverse domains.Source: §1 · pp. 1–2

Approach

  • The study evaluates four diverse datasets under identical conditions using sparse retrieval, fine-tuned transformers, zero-shot LLMs, and leading AVeriTeC 2025 systems.Source: §1 · p. 2

Findings

  • Logistic regression over sparse retrieval outperformed evidence-conditioned LLMs and shared-task top performers because harmful evidence retrieval degraded verification.Source: §1 · p. 2
  • 14-22 points: replacing retrieved evidence with gold annotations improved veracity accuracy across models, showing retrieval quality is a main challenge.Source: §1 · p. 2
  • Macro-F1 0.700 on SciFact fell to 0.315 on ClimateCheck, while AIC CTU and SANCTUARY swapped rankings across datasets and metrics.Source: §1 · p. 2

2 Related Work

Related work established automated fact-checking pipelines, benchmark datasets, retrieval methods, and reproducibility concerns. However, prior benchmarking largely isolated pipeline stages rather than evaluating complete AFC systems across diverse domains.

Automated fact-checking

  • AFC veracity prediction is commonly framed as natural language inference over retrieved evidence, with extensions for graph aggregation, explanation generation, and contrastive robustness.Source: §2 · p. 2
  • Recent agentic systems such as FIRE and DEFAME let the verifier control additional evidence retrieval, moving beyond fixed retrieve-then-verify pipelines.Source: §2 · p. 2

Benchmarks

  • Fact-checking benchmarks span political and open-web claims, scientific claims against biomedical abstracts, climate claims linked to Wikipedia, and live-web evidence retrieval.Source: §2 · p. 2

Evidence retrieval

  • Sparse, dense, and retrieval-augmented methods have advanced evidence retrieval, but specialised training remains necessary for domain-specific AFC corpora.Source: §2 · pp. 2–3

Benchmarking studies

  • Reproducibility studies found classical baselines often remain close to fine-tuned models, while heterogeneous retrieval benchmarks frequently favor BM25 over dense models.Source: §2 · p. 3

3 Tasks, Datasets, and Models

The study evaluates a two-stage AFC pipeline across four datasets organized by evidence domain and claim origin, comparing simple, neural, LLM, and shared-task systems under common metrics.

Tasks

  • Given claim c and corpus D, the pipeline retrieves K relevant documents and predicts Supports, Refutes, or NEI from the claim and retrieved evidence.Source: §3.1 · p. 3

Datasets

  • The four datasets cover open-web AVeriTeC, Wikipedia-based ClimateFEVER, expert-written SciFact, and social-media ClimateCheck claims.Source: §3.2 · p. 3
  • Dataset structure varies along evidence domain and claim origin, with scientific evidence requiring semantic matching and informal claims widening vocabulary gaps.Source: §3.2 · p. 3, Figure 2 · p. 3

Models

  • The evaluation uses random and sparse retrieval, fine-tuned transformers, zero-shot LLMs, and AVeriTeC 2025 winner and runner-up systems.Source: §3.3 · p. 3
  • Claim-only baselines diagnose whether labels can be predicted from claim text without evidence, probing dataset artifacts or annotation shortcuts.Source: §3.3 · pp. 3–4

Evaluation

  • Retrieval is measured with Recall@K and F1@K, while veracity is measured using accuracy and equally weighted macro-averaged F1.Source: §3.4 · p. 4

4 Experimental Results and Analysis

Results across four datasets show unstable system rankings, strong domain effects, and retrieval quality as the main constraint on veracity prediction. Classical and fine-tuned models can outperform larger evidence-conditioned systems when retrieved evidence is noisy or mismatched.

Evidence retrieval

  • Retrieval rankings reverse by domain: AIC CTU reaches R@5 = 0.731 on SciFact versus TF-IDF’s 0.217, while TF-IDF leads AVeriTeC at R@5 = 0.126.TF-IDF consistently outperforms BM25, while ClimateCheck and ClimateFEVER remain difficult for all methods.Source: §4.2 · p. 4

Veracity prediction

  • On ClimateCheck, LONGFORMER achieves the highest accuracy at 0.618, narrowly followed by DISTILROBERTA at 0.617.Retrieved evidence can reduce LLM accuracy under domain shift, and low-quality retrieval can mislead veracity prediction.Source: §4.2 · p. 5
  • On SciFact, SANCTUARY leads with accuracy 0.702 and macro-F1 0.700, while fine-tuned models remain below 0.47 accuracy.SANCTUARY’s sentence-level evidence grouping aligns with SciFact’s rationale-sentence annotations.Source: §4.2 · p. 6

Claim-only signals

  • Claim-only macro-F1 exceeds random prediction by 26.7 points on ClimateCheck, indicating substantial label-predictive information in claim text alone.Weak evidence annotations and annotation ambiguity may contribute to this apparent baseline advantage.Source: §4.2 · p. 6

Retrieval quality

  • 14–22 percentage points: gold evidence improves veracity accuracy across LLM models, confirming retrieval as the primary bottleneck.The largest gains occur for LLAMA 70B (+22 pp) and LLAMA 8B (+20 pp) on ClimateFEVER.Source: §4.3 · p. 7

Cross-dataset rankings

  • No single system dominates across benchmarks: SANCTUARY reaches macro-F1 0.700 on SciFact but 0.315 on ClimateCheck.AIC CTU and SANCTUARY also swap rankings depending on evaluation metrics and datasets.Source: §4.4 · p. 8

5 Conclusion

The study shows that simple baselines remain necessary, while AFC performance varies across domains and metrics and retrieval quality remains central to system performance.

  • The benchmark evaluates evidence retrieval and claim veracity prediction across AVeriTeC, SciFact, ClimateCheck, and ClimateFEVER.Source: §5 · p. 8
  • Claim-only TF-IDF + LOGREG outperform evidence-conditioned zero-shot LLMs and top-performing systems on ClimateCheck.Misleading retrieval can substantially degrade AFC systems.Source: §5 · p. 9
  • No single system dominates across all four benchmarks, with rankings changing by domain and evaluation metric.SANCTUARY spans 0.39 macro-F1 across datasets, while AIC CTU and SANCTUARY swap positions depending on the evaluation setting.Source: §5 · p. 9
  • Replacing retrieved evidence with gold annotations improves accuracy across models and datasets, identifying retrieval as the primary bottleneck.Source: §5 · p. 9
  • Cross-domain AFC evaluation should include classical baselines and explicitly separate retrieval from veracity prediction.Leaderboard rankings should not be treated as proof of real-world utility until systems demonstrate reliability across diverse domains.Source: §5 · p. 9

Limitations

The conclusions are bounded by the study’s four-dataset scope and by limitations in its failure analysis.

  • The four-dataset study may not generalise to all AFC domains or claim types.FEVER and FEVERous were excluded because their scale exceeded the computational budget; multilingual, multi-domain, and newer generative systems were also not evaluated.Source: Limitations · p. 9
  • Failure analysis used two annotators with moderate agreement and covered only a subset of error cases.Source: Limitations · p. 9

Ethical Considerations

The study uses publicly released benchmarks and existing models rather than collecting new data or deploying a production fact-checking system.

  • The evaluation uses AVeriTeC, SciFact, ClimateFEVER, and ClimateCheck, with no newly collected data.Claims and evidence documents come from the original datasets, whose licences permit academic use.Source: Ethical Considerations · p. 9

Generative AI Usage

Generative AI tools were used only to improve the manuscript’s grammar, typos, readability, and written quality.

  • ChatGPT was used to check grammar and typos and enhance the readability and quality of the written text.Source: Generative AI Usage · p. 9

A Evaluation Metrics

The evaluation defines retrieval and veracity metrics for comparing evidence selection and classification under dataset-specific conditions. It also highlights class imbalance, retrieval bias, and annotation ambiguity as important interpretive factors.

Retrieval metrics

  • Recall@K measures coverage of gold evidence, Precision@K measures retrieved-result quality, and F1@K combines both through their harmonic mean.Source: A Evaluation Metrics · p. 13
  • Recall@K plateaus when K exceeds the gold-set size, remaining constant across K ∈{5, 10, 20} for AVeriTeC because its median gold set has one document.Source: A Evaluation Metrics · p. 13
  • BM25’s length normalisation penalises long documents, explaining why it underperforms TF-IDF on long Wikipedia and scientific-abstract corpora.Source: A Evaluation Metrics · p. 13

Veracity metrics

  • Macro F1 weights classes equally, unlike accuracy, and can differ from accuracy by up to 20 points when the majority class reaches 60% of samples.Source: A Evaluation Metrics · p. 13

Metric choice

Dataset and result interpretation

  • Class imbalance drives stronger performance on dominant labels and near-zero performance on rare labels, especially Conflicting Evidence, while ClimateCheck NEI is learned more reliably by transformers.Source: §B.3 · p. 15
  • Moderate agreement and annotation judgments complicate small score differences: annotators agreed at κ = 0.55 and classified 38% of sampled errors as annotation mistakes or debatable cases.Source: §E.1 · p. 17
  • Longer documents are retrieved more often on SciFact and ClimateCheck because claim-adjacent tokens inflate Recall@K independently of true relevance.Source: §E.1 · p. 18

E.2 Per-Dataset Analysis

Per-dataset errors arise from evidence fragmentation, lexical mismatch, multi-passage requirements, label ambiguity, and dataset-specific annotation schemes. Cross-dataset results show that Conflicting Evidence is broadly difficult and annotation issues limit interpretation of small margins.

AVeriTeC

  • AVeriTeC retrieval often returns evidence fragments that are individually insufficient to support or refute claims.Source: §E.2 · p. 18
  • SANCTUARY over-predicts Refuted on AVeriTeC, while Conflicting Evidence remains nearly unpredictable with F1 < 0.18 for all models.Source: §E.2 · p. 18

SciFact

  • SciFact failures combine lexical mismatch with sentence-level rationale granularity, causing retrievers to miss semantically related evidence and verifiers to predict NEI.Source: §E.2 · p. 18

ClimateCheck

  • ClimateCheck has the widest vocabulary gap: informal social-media claims and formal scientific abstracts frequently share few tokens, producing irrelevant retrieval.Source: §E.2 · p. 18

ClimateFEVER

  • ClimateFEVER typically requires 2–5 Wikipedia passages, so single-passage retrieval can be insufficient; this yields the study’s largest oracle gains of +20–22 pp.Source: §E.2 · p. 18

Cross-dataset patterns

  • Across datasets, Conflicting Evidence scores 0.00–0.20, LLMs often over-predict NEI, and annotation issues make margins below 5 accuracy points unreliable.Source: §E.2 · p. 19

F Frontier-LLM Verifier Experiments

Frontier-LLM experiments isolate evidence quality by holding the verifier and prompt fixed, then compare claim-only, retrieved, gold, and iterative retrieval conditions. Gold evidence performs best, while retrieval quality constrains both one-shot and agentic systems.

Experimental design

Three evidence conditions

  • Gold evidence produces the best AVeriTeC accuracy at 87.3, compared with 81.0 for claim-only input and 38.1 for retrieved evidence.Source: §F.1 · p. 21
  • A 49-point AVeriTeC gap between gold and retrieved evidence is attributable to evidence quality alone under the fixed-model comparison.Source: §F.1 · p. 21

Iterative retrieval

  • Agentic retrieval remains below oracle performance—87.3 on AVeriTeC and 70.9 on ClimateCheck—with second rounds triggered for 25.4% and 30.9% of claims.Source: §F.2 · p. 21, Table 15 · p. 21

G Recommendations for AFC Evaluation Design

The paper recommends evaluation practices that expose baseline competitiveness, domain transfer, retrieval limits, class-specific behavior, and annotation quality. These practices are intended to make AFC comparisons more reliable and interpretable.

Baseline and scope

Pipeline diagnosis

Metric reporting

Annotation quality

Related papers

Loading 2608.25934v1…