Evidence-Grounded Multimodal Misinformation Detection with Attention-Based GNNs

Abhishek Tyagi; Adiba Mahbub Proma; Mir Nafis Sharear Shopnil; Sharad Duwal

REVIEW 4 major objections 5 minor 1 cited by

Evidence-Grounded Multimodal Misinformation Detection with Attention-Based GNNs

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A graph attention network that compares a caption's claim graph with an evidence graph built from reverse-image-search web pages outperforms frontier LLMs at out-of-context misinformation detection on the Factify benchmark.

desk verdict A useful baseline and new datasets, but the headline accuracy is measured on a cherry-picked LLM-abstention subset, not the full evaluation set. read the letter →

arxiv 2505.18221 v1 pith:25HUHWY2 submitted 2025-05-23 cs.LG cs.AIcs.CLcs.IR

Sharad Duwal , Mir Nafis Sharear Shopnil , Abhishek Tyagi , Adiba Mahbub Proma This is my paper

classification cs.LGcs.AIcs.CLcs.IR

keywords out-of-contextmisinformationmultimodaldetectiongraphneuralnetworkscross-graphattentionevidencegraphsreverseimagesearchFactifydatasetLLMcomparison

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to show that out-of-context misinformation—real images paired with false captions—can be detected by grounding the image in online textual evidence and comparing two graphs rather than by relying on an LLM's memory. The authors build an evidence graph from the top-ranked web pages that reuse the image and a claim graph from the caption, then train a small graph attention network to score whether the caption is consistent with the evidence. On the Factify validation subset the method reaches 93.05% accuracy on the 461 samples all LLMs could answer and 83.05% on all 1,145 samples, beating the best LLM by 2.82 points on the common set with about 10 million parameters. If this holds, task-specific graph detectors are a cheaper, less hallucination-prone alternative to LLM-based fact-checking.

What carries the argument

The load-bearing object is the pair of graphs produced by the EGMMG pipeline. A reverse image search collects web pages containing the image; after ranking by CLIP-based cosine similarity between page text and image, the top 7 pages are concatenated into the evidence. A dependency parser extracts entities, events, and locations as nodes, with rule-based edges such as PERFORMS, EXPERIENCES, TARGETS, LOCATED_IN, HAS_STATE, and SAME_AS derived from verb and dependency patterns; the caption is turned into the claim graph the same way. The classifier initializes node features as a learnable weighted sum of BERT label embeddings and neighborhood structural features, passes messages with transformer-style graph convolution layers, scores node importance, and computes cross-attention with claim nodes as queries and evidence nodes as keys and values. Global mean pooling concatenates evidence, claim, and attended representations into a sigmoid veracity score, so the model's decision depends directly on the consistency between the two graphs.

What would settle it

Take the 1,855 Factify validation samples the pipeline discarded for lack of web matches, obtain ground-truth labels, build graphs for any that have partial matches, and run EGMMG: if accuracy on that group is near chance, the reported 83.05% to 93.05% results only characterize images with online evidence. Alternatively, have annotators judge whether the top-7 retrieved pages for a random sample of 100 pairs actually discuss the entities and event named in the caption; if most are irrelevant, the evidence graphs do not carry the context the method relies on.

Watch

Extended reading notes

Core claim

The central claim is that the relative agreement between the claim graph and the evidence graph carries the veracity signal, and that a graph attention model reading these two graphs can outperform frontier LLMs given the same retrieved evidence. The paper reports 0.9305 accuracy and 0.9219 F1 on EVAL_COMMON and 0.8305 accuracy and 0.8455 F1 on EVAL_ALL, compared with GPT-4o's 0.9023 and 0.8936 on EVAL_COMMON and 0.6872 and 0.6209 on EVAL_ALL, plus cross-dataset results of 0.8248 on Factify, 0.7750 on COSMOS, and 0.7100/0.6823 on MMFakeBench under 85:15 splits. The evidence graph supplies the context that the caption must be checked against, while the claim graph encodes the caption's own entities and relations. On the paper's account, the advantage comes from doing this contextualization explicitly instead of relying on parametric knowledge.

Load-bearing premise

The method assumes that the web pages returned by reverse image search describe the true context of the image, and that ranking them by CLIP similarity and keeping the top 7 preserves the facts needed to judge the caption; it also assumes the images dropped for lack of matches are no different from the ones kept.

Editorial extensions

If this is right

EGMMG's 93.05% accuracy on EVAL_COMMON and 83.05% on EVAL_ALL imply that explicit evidence grounding can beat zero-shot LLM prompting on the same evidence, at least on this benchmark.
At 10.7 million parameters running on a single T4, the approach is cheap enough to deploy at scale where API-based LLM fact-checking would be too costly.
The cross-dataset results indicate the pipeline transfers without per-dataset engineering, though with weaker performance on the more synthetic MMFakeBench sets than on Factify and COSMOS.
Ablation results show that weighted node features and 768-dimensional label embeddings matter most, while edge features reduce accuracy, pointing to node semantics rather than edge types as the informative part of the graph.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The gap between 93.05% on the 461-sample common set and 83.05% on all 1,145 samples suggests the comparison with LLMs is most favorable on samples every LLM found answerable; deployment on arbitrary image-caption pairs may see the lower number.
Because images without any reverse-image-search match are dropped (Factify shrinks from 14,000 to 4,945 training samples and 3,000 to 1,145 validation samples), the reported accuracy may not hold for social-media posts whose images are not indexed online, and a selection-bias check on discarded samples would tell.
A direct testable extension is to feed the same top-7 evidence documents to an LLM in plain text and compare with the graph encoding, isolating whether the graph structure or the retrieval step is responsible for the gain.
If the evidence premise is right, adding image-derived entities to the evidence graph could recover context for the discarded no-match cases, since the paper notes it currently does not use the image content directly.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

Here's my read on arXiv:2505.18221. The core contribution is a plausible evidence-grounded pipeline for OOC detection: reverse image search, evidence and claim graphs, attention-based GNN with cross-attention. The authors build new graph datasets from Factify, COSMOS, and MMFB, and report ablations showing component contributions. The 10M-parameter model is efficient, and the paper is honest in its limitations section.

The problem is the headline. The abstract claims 93.05% accuracy on 'the evaluation set' and outperforming the second-best LLM by 2.82%. Table 2 shows that number is on EVAL_COMMON (n=461), a subset where all zero-shot LLMs agreed they had enough evidence and answered. On the full EVAL_ALL (n=1145), accuracy is 83.05% — still 14 points above GPT-4o, but that's not what the abstract says. EVAL_COMMON is not a random holdout; it selects for cases where the evidence retrieval already produced text LLMs could use. EGMMG never abstains and is supervised, so the comparison there is not apples-to-apples.

There are also smaller issues: no confidence intervals or multiple seeds for the main table; the LLM baselines get only text, not the image, so the claim that 'all methods have access to the same data' is misleading; and the assumption that reverse image search returns relevant context is load-bearing, with a large chunk of images discarded (Factify drops from 14,000 to 4,945). The authors acknowledge these limitations, which is to their credit, but the evaluation design still overstates the result.

What's genuinely useful: the graph construction rules and the resulting datasets could be reused, and the approach is a legitimate baseline for the 'contextualize first' line of work. The paper deserves a serious referee, but I'd ask for the headline to be corrected to the full-set number, plus variance estimates and a supervised baseline comparison before acceptance. Would I cite it? Not in my next 12 months, but I'd point students to it as a solid baseline example.

Referee Report

4 major / 5 minor

Summary. The paper proposes EGMMG, a GNN-based detector for out-of-context multimodal misinformation. It first retrieves web evidence for an image via reverse image search, filters the retrieved documents by CLIP similarity, constructs entity-relation graphs from the evidence text and from the caption, and trains an attention-based GNN to classify image-caption pairs. Experiments on Factify, COSMOS, and MMFakeBench compare the method with zero-shot LLMs; the paper reports 93.05% accuracy on the EVAL_COMMON subset (n=461) and 83.05% on EVAL_ALL (n=1145), with ablations and generalization tests on additional datasets.

Significance. If the result holds, the paper is a useful demonstration that a small task-specific GNN fed with retrieved textual evidence can compete with frontier LLMs for out-of-context misinformation detection, and the evidence-hydration pipeline is a reusable resource. The strengths are the use of multiple public datasets, a clearly described retrieval-and-graph construction pipeline, and an honest limitations section. However, the headline accuracy is reported on a post hoc subset defined by LLM abstention, no confidence intervals or multiple runs are given for the main table, and the comparison lacks a trained non-graph supervised baseline, so the significance as stated is not yet fully established.

major comments (4)

[Abstract, Section 4.1, Table 2] The headline claim of '93.05% detection accuracy on the evaluation set' is read from the EVAL_COMMON column (n=461), but EVAL_COMMON is not the evaluation set; it is the subset of EVAL_ALL (n=1145) on which the zero-shot LLMs did not abstain. Because abstention is triggered by insufficient or ambiguous evidence, EVAL_COMMON is selected toward samples that the LLMs already judged answerable, and EGMMG never abstains. The full-set accuracy is 83.05% on EVAL_ALL, which still beats GPT-4o by a large margin, so the central direction may survive, but the abstract's specific number and the 2.82% margin are not the evaluation-set result. Please report EVAL_ALL as the primary evaluation set and present EVAL_COMMON only with an explicit statement of the selection mechanism and its implications.
[Section 4.1, EVAL_COMMON definition] The definition of EVAL_COMMON lists ESSonnet ∩ ESGpt4oMini ∩ ESHaiku, omitting GPT-4o, while the surrounding text says the set contains samples that all models considered answerable. This matters because GPT-4o is the strongest LLM baseline and the one closest to EGMMG on EVAL_COMMON. In addition, Table 2 reports no confidence intervals, bootstrap estimates, or multiple training seeds for the main comparison; with n=461, the 2.82% accuracy margin corresponds to roughly 13 samples, so a significance test or interval is needed before claiming a reliable margin over GPT-4o.
[Section 3.1, Table 1] The evidence-retrieval filter removes a large fraction of each dataset, e.g., Factify drops from 14,000 to 4,945 samples and the Factify validation set from 3,000 to 1,145. The paper does not check whether retained and discarded samples are exchangeable in label distribution, claim type, or image properties. If successful reverse-image search correlates with the nature of the claim or with veracity, then the evaluation is on a biased subpopulation and the generalization claims in Section 4.2 are not supported. Please report class balance and available content statistics before and after filtering, or justify why selection bias is not a concern.
[Sections 4, 5, and 6] The comparison is between EGMMG, which is trained on the dataset labels, and LLMs used in a zero-shot setting, so the experiment does not isolate the graph-based architecture as the source of the observed gain. A supervised non-graph baseline (e.g., a fine-tuned text classifier over the same retrieved evidence) is needed to support the claim that the graph structure itself is responsible for the improvement. Relatedly, Section 6's statement that 'all methods have access to the same amount of data (in text or graph format)' is not literally accurate: LLMs receive raw evidence text and the claim, EGMMG receives a lossy graph projection of the evidence text, and neither receives the image directly. Please add the missing supervised text baseline or temper the architecture-specific claim.

minor comments (5)

[Section 4.1] There are typos: 'Facitify' should be 'Factify', and 'postitive' should be 'positive'.
[Section 3.1] The text introduces 'top-k evidences' and then fixes k=7; please make the notation consistent and state whether any sensitivity analysis was performed on the choice of k.
[Table 6] The number of runs differs between the 384-dim setting (four runs) and the 768-dim setting (three runs); please state why and report mean ± standard deviation instead of raw run lists.
[Tables 2 and 3] Please clarify the relationship between the Factify split used to train the model for Table 2 (EVAL_ALL/EVAL_COMMON) and the 'Factify 85:15' split used for Tables 3 and 6, since both report accuracy on Factify-derived data.
[Figure 5 and Section 4.1] For EVAL_ALL, Sonnet abstained on 64 samples even when prompted to answer strictly; the paper should state explicitly how abstained outputs are treated in the reported accuracy and F1 scores.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the method is an empirically trained supervised system whose graph construction is label-independent; the headline 93.05% is a conditional subset result, not a derivation forced by construction.

full rationale

I walked the paper's claimed derivation chain. The evidence pipeline in Section 3.1 builds evidence and claim graphs using reverse image search, CLIP cosine similarity ranking, and spaCy rule-based entity/relation extraction; none of these steps consume the target label, so the graph data are not constructed from the supervision signal. The classifier in Section 3.2 is trained in a standard supervised loop on Factify-derived graph data, and the reported accuracies are empirical measurements on held-out subsets, not quantities obtained by plugging the labels into the model's own definitions. The comparison against LLMs is a shared-input comparison in the intended sense: both EGMMG and the LLMs receive the same retrieved evidence text (EGMMG as entity-relation graphs, the LLMs as raw text in the Figure 5 prompt), and the paper's own Limitations section admits that the method does not use the image directly, consistent with the LLM prompt also omitting the image. The only respect in which the headline number is vulnerable is that '93.05% on the evaluation set' is read from EVAL_COMMON (n=461), a subset defined by the zero-shot LLMs' abstention behavior; on EVAL_ALL (n=1145) the method scores 83.05%. That is a selective-reporting and generalization concern about what 'the evaluation set' denotes, not a circular reduction: EGMMG's accuracy on EVAL_COMMON is not forced by the subset's definition, and no parameter is fitted to the EVAL_COMMON labels. There are no load-bearing self-citations, no imported uniqueness theorem, no ansatz smuggled via citation, and no renaming of a known result as a new derivation. The central empirical claim is therefore not equivalent to its inputs by construction; the paper's circularity is negligible, with the important caveat that the abstract's headline number is conditional on a favorable evaluation subset.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central result rests on the reliability of online evidence retrieval, the sufficiency of a lossy graph projection, and the exchangeability of discarded samples. No new physical or conceptual entities are introduced.

free parameters (2)

top-k evidence documents = 7
The pipeline selects the top 7 evidence documents after CLIP cosine similarity ranking; no ablation over k is reported, and the central result depends on this choice.
model hyperparameters = hidden dim 1024; 2 conv layers; lr 3e-4; batch size 64
These hand-chosen hyperparameters affect the reported accuracy; the paper does not study sensitivity to them.

assumptions (3)

domain assumption Reverse image search returns web pages that provide the true context of the image.
Used in Section 3.1 to build evidence graphs; if the retrieved pages are irrelevant or misleading, the graph comparison cannot establish veracity.
domain assumption The spaCy-based entity-relation graph projection preserves the information needed to assess claim-image consistency.
The graph construction in Appendix B reduces raw text to six edge types; if this lossy projection drops key facts, the GNN cannot recover them.
domain assumption Discarding images without web matches does not bias the evaluation.
Table 1 shows large drops in dataset size (e.g., Factify from 14,000 to 4,945); the paper does not compare retained and discarded samples.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Evidence-Grounded Multimodal Misinformation Detection with Attention-Based GNNs." pith.science (2026). https://pith.science/paper/25HUHWY2

@misc{pith2026250518221,
  author       = {Pith},
  title        = {Pith review of: Evidence-Grounded Multimodal Misinformation Detection with Attention-Based GNNs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/25HUHWY2}},
  note         = {Machine review of arXiv:2505.18221}
}

abstract

Multimodal out-of-context (OOC) misinformation is misinformation that repurposes real images with unrelated or misleading captions. Detecting such misinformation is challenging because it requires resolving the context of the claim before checking for misinformation. Many current methods, including LLMs and LVLMs, do not perform this contextualization step. LLMs hallucinate in absence of context or parametric knowledge. In this work, we propose a graph-based method that evaluates the consistency between the image and the caption by constructing two graph representations: an evidence graph, derived from online textual evidence, and a claim graph, from the claim in the caption. Using graph neural networks (GNNs) to encode and compare these representations, our framework then evaluates the truthfulness of image-caption pairs. We create datasets for our graph-based method, evaluate and compare our baseline model against popular LLMs on the misinformation detection task. Our method scores $93.05\%$ detection accuracy on the evaluation set and outperforms the second-best performing method (an LLM) by $2.82\%$, making a case for smaller and task-specific methods.

Figures

Figures reproduced from arXiv: 2505.18221 by the authors.

**Figure 1.** The EGMMG pipeline. For an image-claim sample, the pipeline prepares two graphs, evidence graph and claim graph, using online evidence retrieval followed by a rule-based analysis of subject-object relations in the evidence documents. Once we have the two graphs, we use a graph attention-based classifier to detect misinformation. Images repurposed with different captions to make false claims are categorized as out-o… view at source ↗

**Figure 2.** Data sample: Image and claim graph for claim «Sonia Gandhi, the interim Congress president, cast her vote at Nirman Bhawan in the New Delhi assembly constituency, accompanied by Priyanka Gandhi Vadra, who also voted at a booth in Lodhi Estate.» A section of the evidence graph is provided in the Appendix ( [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** The EGMMG classifier. 3.2.3 Architecture After initializing the node and edge features for both the evidence and claim graphs as described above, we perform message-passing between the nodes using graph convolutions (GATConv (Velickovi ˇ c et al. ´ , 2018), TransformerConv (Shi et al., 2021), GATv2Conv (Brody et al., 2022)) to update the node representations. This allows subgraph neighborhoods to inform each node… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Evidence graph generated by EGMMG for the example in Figure [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

**Figure 5.** Figure 5: Prompt used to evaluate misinformation detection performance of LLMs (Sonnet, Haiku, GPT). For the [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Multimedia Verification Through Multi-Agent Deep Research Multimodal Large Language Models
cs.CV 2025-07 conditional novelty 4.0 of 10

A six-stage multi-agent MLLM pipeline with reverse image search, metadata analysis, and fact-checking tools is demonstrated on a single Ukraine missile-strike video, with no quantitative evaluation.

Reference graph

Works this paper leans on

13 extracted references · 12 canonical work pages · cited by 1 Pith paper

[1]

Carefully read the evidence document, which consists of excerpts from multiple news articles

work page
[2]

Analyze the claim provided and compare it to the evidence

work page
[3]

true" or

Respond with "true" or "false" based on your analysis. Do not provide explanations or additional commentary. EVIDENCE: {evidence} CLAIM: {claim} Your response should be exactly one of: TRUE, FALSE. YOUR RESPONSE: Figure 5: Prompt used to evaluate misinformation detection performance of LLMs (Sonnet, Haiku, GPT). For the EVAL_SUFFICIENTset, we allow one mo...

work page arXiv
[7]

InPro- ceedings of the Twelfth Language Resources and Evaluation Conference, pages 6149–6157, Marseille, France

Fakeddit: A new multimodal benchmark dataset for fine-grained fake news detection. InPro- ceedings of the Twelfth Language Resources and Evaluation Conference, pages 6149–6157, Marseille, France. European Language Resources Association. Tobias A. Opsahl. 2024. Fact or fiction? improving fact verification with knowledge graphs through simpli- fied subgraph...

work page 2024
[8]

image, tell me your story!

Sniffer: Multimodal large language model for explainable out-of-context misinformation detection. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), pages 13052–13062. Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. Nils ...

work page 2019
[2016]

Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, and Haofen Wang

Pizzagate: From rumor, to hashtag, to gunfire in d.c. Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, and Haofen Wang. 2024. Retrieval-augmented gen- eration for large language models: A survey. Matthew Honnibal, Ines Montani, Sofie Van Lan- deghem, and Adriane Boyd. 2020. spaCy: Industrial- strength Na...

work page 2024
[2018]

the", we additionally map

Graph attention networks. Haoran Wang and Kai Shu. 2023. Explainable claim verification via knowledge-grounded reasoning with large language models. InFindings of the Associa- tion for Computational Linguistics: EMNLP 2023, pages 6288–6304, Singapore. Association for Com- putational Linguistics. Keyang Xuan, Li Yi, Fan Yang, Ruochen Wu, Yi R. Fung, and He...

work page 2023
[2020]

Canyu Chen and Kai Shu

Language models are few-shot learners. Canyu Chen and Kai Shu. 2024. Can LLM-generated misinformation be detected? InThe Twelfth Interna- tional Conference on Learning Representations. Emily Denniss and Rebecca Lindberg. 2025. Social media and the spread of misinformation: infectious and a threat to public health.Health Promotion In- ternational, 40(2):da...

work page 2024

Show all 13 references

[2021]

Jiho Kim, Sungjin Park, Yeonsu Kwon, Yohan Jo, James Thorne, and Edward Choi

Ia-gcn: Interpretable attention based graph convolutional network for disease prediction. Jiho Kim, Sungjin Park, Yeonsu Kwon, Yohan Jo, James Thorne, and Edward Choi. 2023. Factkg: Fact verifi- cation via reasoning on knowledge graphs. Thomas N. Kipf and Max Welling. 2017. Se...

2023 arXiv
[2022]

InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14940–14949

Open-domain, content-based, multi-modal fact-checking of out-of-context images via online re- sources. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14940–14949. Shivangi Aneja, Chris Bregler, and Matthias Nießner
[2023]

InProceedings of the AAAI conference on artificial intelligence, vol- ume 37, pages 14084–14092

Cosmos: catching out-of-context image mis- use using self-supervised learning. InProceedings of the AAAI conference on artificial intelligence, vol- ume 37, pages 14084–14092. Shaked Brody, Uri Alon, and Eran Yahav. 2022. How attentive are graph attention networks? Tom B. Brow...

2022
[2024]

Lisa Fazio

Ammeba: A large-scale survey and dataset of media-based misinformation in-the-wild. Lisa Fazio. 2020. Out-of-context photos are a powerful low-tech form of misinformation.The Conversation, 14(1). Marc Fisher, John Woodrow Cox, and Peter Hermann

2020
[2025]

Shaydanay Urbani

Cove: Context and veracity prediction for out-of-context images. Shaydanay Urbani. 2020. Verifying online information. Petar Veliˇckovi´c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio

2020

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Carefully read the evidence document, which consists of excerpts from multiple news articles

work page

[2] [2]

Analyze the claim provided and compare it to the evidence

work page

[3] [3]

true" or

Respond with "true" or "false" based on your analysis. Do not provide explanations or additional commentary. EVIDENCE: {evidence} CLAIM: {claim} Your response should be exactly one of: TRUE, FALSE. YOUR RESPONSE: Figure 5: Prompt used to evaluate misinformation detection performance of LLMs (Sonnet, Haiku, GPT). For the EVAL_SUFFICIENTset, we allow one mo...

work page arXiv

[4] [7]

InPro- ceedings of the Twelfth Language Resources and Evaluation Conference, pages 6149–6157, Marseille, France

Fakeddit: A new multimodal benchmark dataset for fine-grained fake news detection. InPro- ceedings of the Twelfth Language Resources and Evaluation Conference, pages 6149–6157, Marseille, France. European Language Resources Association. Tobias A. Opsahl. 2024. Fact or fiction? improving fact verification with knowledge graphs through simpli- fied subgraph...

work page 2024

[5] [8]

image, tell me your story!

Sniffer: Multimodal large language model for explainable out-of-context misinformation detection. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), pages 13052–13062. Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. Nils ...

work page 2019

[6] [2016]

Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, and Haofen Wang

Pizzagate: From rumor, to hashtag, to gunfire in d.c. Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, and Haofen Wang. 2024. Retrieval-augmented gen- eration for large language models: A survey. Matthew Honnibal, Ines Montani, Sofie Van Lan- deghem, and Adriane Boyd. 2020. spaCy: Industrial- strength Na...

work page 2024

[7] [2018]

the", we additionally map

Graph attention networks. Haoran Wang and Kai Shu. 2023. Explainable claim verification via knowledge-grounded reasoning with large language models. InFindings of the Associa- tion for Computational Linguistics: EMNLP 2023, pages 6288–6304, Singapore. Association for Com- putational Linguistics. Keyang Xuan, Li Yi, Fan Yang, Ruochen Wu, Yi R. Fung, and He...

work page 2023

[8] [2020]

Canyu Chen and Kai Shu

Language models are few-shot learners. Canyu Chen and Kai Shu. 2024. Can LLM-generated misinformation be detected? InThe Twelfth Interna- tional Conference on Learning Representations. Emily Denniss and Rebecca Lindberg. 2025. Social media and the spread of misinformation: infectious and a threat to public health.Health Promotion In- ternational, 40(2):da...

work page 2024