Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection

Beizhe Hu; Danding Wang; Juan Cao; Qiang Sheng; Zhengjia Wang

REVIEW 3 major objections 5 minor 22 references

Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read InSide proposes that modeling the intent behind news, together with its semantic graph, detects fake news more reliably than semantic-only or fixed-intent systems, and reports consistent gains on four benchmarks.

desk verdict Solid graph-based detector with consistent benchmark gains; the 'stable intent' story needs validation before it carries the weight the paper puts on it. read the letter →

arxiv 2509.01660 v1 pith:R5DJ74B2 submitted 2025-09-01 cs.CL

Zhengjia Wang , Qiang Sheng , Danding Wang , Beizhe Hu , Juan Cao This is my paper · ORCID

classification cs.CL

keywords fakenewsdetectionintentmodelingintent-semanticjointlearningheterogeneousgraphneuralnetworksalignmentlargelanguagemodelextractionsocialmediamisinformationtemporalrobustness

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

The paper tries to show that a news item's intent—the purpose or motivation behind writing it—is a durable clue for fake news detection, even when the surface wording shifts. It claims that intent can be extracted from content with a frozen large language model and combined with sentence- and entity-level semantics through graph-based joint learning, and that this combination is more robust than semantic-only or fixed-category intent models. On PolitiFact, GossipCop, Weibo, and LLMFake, the proposed InSide model reports consistently higher macro-F1 and fake-class F1 than the compared baselines, with the largest fake-class improvement about 9.12% on GossipCop. The reason this would matter: a content-only detector that tracks stable intent could catch newly published fake news before comments, propagation structure, or fact checks become available.

What carries the argument

The carrier of the argument is a pair of heterogeneous graphs—one for semantics (sentences plus entities with sliding-window and entity edges) and one for intent (LLM-derived coarse nodes plus learnable fine-grained nodes)—joined by dynamic pathway-based graph alignment through pseudo nodes. The pseudo nodes create a common space with edges to both graphs, use attention over typed edge features to decide which pathways matter, and are mean-pooled into the final classifier, so they act as the information bottleneck that fuses 'thoughts' and 'words'.

What would settle it

Run InSide against an ablated control in which each LLM-generated intent string is replaced by an equally long string formed from random sentences of the same news article, keeping every other component identical. If macro-F1 stays near 0.86 on GossipCop, intent content is not the operative signal. A complementary check: give human annotators the extracted intent strings without the news text and ask them to label fake versus real; chance-level performance would mean intent alone does not separate the classes.

Watch

Extended reading notes

Core claim

At its core, the paper claims that writing style is flexible but intent is comparatively stable: two deceptive articles can sound very different yet share the same underlying intent, so intent offers a reliable target for detection. To use that idea, InSide converts one article into two heterogeneous graphs. The semantic graph has sentence nodes connected by a sliding window plus entity nodes that create long-range links; the intent graph has coarse nodes representing belief, plan, desire, and outcome, produced by prompting a frozen large language model, plus learnable fine-grained nodes that tie each coarse intent to the narrative. Both graphs are updated by local message passing and a supe

Load-bearing premise

The load-bearing premise is that the texts a frozen large language model produces when prompted for belief, plan, desire, and outcome really are a faithful, non-redundant record of the news's intent; if those texts are just paraphrases of the article's surface meaning, the reported gains could come from extra model capacity or LLM priors rather than from intent itself.

Editorial extensions

If this is right

Content-only detection can be made more practical for brand-new articles, because InSide needs no user comments, propagation trees, or external evidence at inference time.
Intent-aware modeling can be decoupled from any specific taxonomy: InSide with a nine-perspective intent framework stays competitive, suggesting the coarse-to-fine machinery, not the particular prompt set, carries much of the gain.
The joint framework scales with the backbone: replacing BERT with BERT-large or the intent generator with a 32B-parameter model improves results, while lighter modules still help.
Fake-class F1 is the practically important metric for moderation, and relative gains up to 9.12% on GossipCop mean fewer fake items slip through at the same operating point.

Reading between the lines

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

A direct experiment the paper does not run: adversarially rewrite fake news to change style while preserving intent; if InSide degrades less than semantic-only baselines, the 'intent is stable' mechanism is confirmed rather than just the architecture.
The framework could transfer to detecting manipulative framing of true stories, which the paper notes as a possibility, suggesting a broader early-warning use case for coordinated influence campaigns that reuse the same intent across languages.
Because coarse intent nodes come from a frozen LLM, quantizing or distilling the generator could lower deployment cost; whether a fine-tuned intent extractor would outperform the frozen one remains an open question.

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

Quick take: InSide is a real architectural contribution and the empirical package is mostly clean. The headline claim—jointly modeling semantics and LLM-extracted intent as heterogeneous graphs with pseudo-node alignment beats strong baselines—is internally consistent: gains on all four datasets, paired t-tests, and ablations where every removed component hurts. The framework-swap robustness test is a nice touch. Credit where due: this is not a toy experiment.

The biggest soft spot is the intent construct. The paper never validates that the LLM's coarse intent outputs are anything more than rephrased surface text. The swap test in 4.3.3 only changes the taxonomy, not the underlying fidelity problem. An ablation that replaces coarse intents with a paraphrase of the article (or with random text) would tell you whether the gains come from intent, from extra model capacity, or from LLM priors. Without that, the 'stable intent' mechanism is a hypothesis, not a demonstrated result. The benchmark numbers can still be right; the explanation is unproven.

Reproducibility is the second issue. No code or data, several hyperparameters left implicit, and the tuning protocol isn't described. The LLM baselines are nearly degenerate on PolitiFact (macF1 around 0.41), so the claim of beating LLM baselines is weak there. LLMFake's random split also muddies the temporal-generalization narrative, though the paper does disclose it.

None of this is fatal. The paper is honest about its limitations and the empirical claims are specific and falsifiable. If the code and prompts ship, the table of results will be checkable. For a reader in fake news detection, this is worth a serious look; the architecture is novel and the benchmark results are sufficiently strong that ignoring them would be a mistake.

My recommendation: send it to review, but require the missing intent-fidelity analysis and the code/data release before acceptance. The central mechanism claim should be reworded as a hypothesis supported by ablations, not a validated finding.

Referee Report

3 major / 5 minor

Summary. The paper proposes InSide, a graph-based joint modeling framework for fake news detection that combines semantic signals (sentence and entity graphs) with LLM-extracted intent signals (coarse and fine-grained intent nodes). The method builds two heterogeneous graphs, applies dual-level local/global message passing, and introduces a dynamic pathway-based alignment module with pseudo nodes to bridge the semantic-intent representation gap. Experiments on PolitiFact, GossipCop, Weibo, and LLMFake report consistent improvement over intent-agnostic and intent-aware baselines on accuracy, macro-F1, AUC, and class-wise F1. The central conceptual claim is that news intent is a stable deception signal even when surface writing patterns shift, and that InSide exploits this stability through joint intent-semantic learning.

Significance. If the empirical results hold, InSide is a substantive advance for content-based fake news detection: it demonstrates that LLM-generated intent signals can be integrated with semantic graph structures and yields consistent gains across four datasets, with ablations showing each component contributes. The paper's strengths include evaluation under chronological splits (except LLMFake), comparison against strong intent-aware baselines, an extensibility study across semantic encoders and intent generators, and a case visualization of learned alignment pathways. The main weakness is that the 'intent' construct is not independently validated; the reported gains may stem from additional model capacity or LLM priors rather than from the hypothesized stable-intent mechanism. This gap is central to the paper's motivation and limits the interpretability of the results.

major comments (3)

[Section 3.3 and Section 4.3.3] The load-bearing conceptual claim is that LLM-extracted coarse intents capture stable intent rather than surface paraphrase. However, no human validation of the generated intents is provided, and the framework-swap test InSide(C') only changes the taxonomy used in the prompt; it does not test whether the coarse nodes encode content distinct from a simple re-encoding of the article. To support the central mechanism, please add (i) human evaluation of a sample of generated intents, or (ii) a control ablation that replaces coarse intent nodes with surface-level LLM summaries or paraphrases of the news text. Without such a control, the performance gains could be attributed to extra capacity or to implicit veracity priors in the frozen LLM.
[Section 3.3, fine-grained intent initialization] The fine-grained intent node update (hf_ij <- hf_ij + softmax((hc_i + hf_ij) · Hsem_sen^T) · Hsem_sen) directly mixes coarse intent embeddings with sentence embeddings. Thus the 'intent' graph is not independent of semantics, and the w/o Intf ablation removes both the fine nodes and a substantial amount of semantic-conditioned parameters. The improvement might reflect increased capacity rather than intent-specific information. Please include an ablation that replaces fine-grained intent nodes with an equivalent-capacity semantic-only component, or otherwise controls for parameter count, so that the intent contribution is isolated.
[Section 4.2, Table 2] The asterisk is defined as a paired t-test at 0.05 significance comparing InSide with the best baseline, but only three runs are reported. Please clarify whether the paired test is computed over test instances (which ignores run-level variability) or over the three runs (which would be a very small sample). Additionally, because five metrics across four datasets are tested, multiple-comparison correction or at least a conservative interpretation should be reported. This is important because the paper's central claim—'consistently outperforms all competitive methods'—relies entirely on this significance evidence.

minor comments (5)

[Section 3.3] Typo: 'denoted as Lcoarse and Lcoarse' should be 'Lcoarse and Lfine'. Also, the sentence 'Lfine encodes logical dependencies among high-level intent aspects' seems to describe coarse-coarse edges; please clarify the edge semantics.
[Section 4.2] Typo: 'the feature extractor of MEInt' should be 'DMInt'.
[Table 3 and Figure 3] Minor typos: 'PoltiFact' in Table 3 header and 'Weibo21' in Figure 3 axis labels; use consistent dataset names.
[Reproducibility] No code or data availability statement is provided. Given the reliance on LLM-generated intents and specific hyperparameters, releasing code and the exact prompts would materially help reproducibility.
[Prompt 1] The prompt text says the underlined part is only for LLM-Int, but the underline is not visible in the manuscript. Please mark it explicitly (e.g., with brackets) so readers can distinguish the two variants.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: benchmark-anchored evaluation and framework-agnostic intent extraction keep the derivation self-contained.

full rationale

The paper's central empirical claim is that InSide outperforms existing methods on four external benchmark datasets (PolitiFact, GossipCop, Weibo, LLMFake) against baselines with published settings. This is not circular: the performance numbers are not derived from the method's assumptions, and no parameter fitted on a subset is renamed as a prediction. The intent representation is produced by a frozen LLM prompted with a belief/plan/desire/outcome taxonomy from Wang et al. (2025), which shares authors with this paper. However, the taxonomy is only one possible source of coarse-grained intent nodes, and the paper explicitly tests an alternative framework in the InSide(C') experiment (Section 4.3.3), showing stable performance. Thus the self-citation supplies motivation and a default prompt, not a load-bearing uniqueness or impossibility result. The conceptual claim that 'intent is stable while semantics shift' is not directly validated by human annotation or paraphrase controls, but that is a correctness/validity concern about whether the LLM outputs capture intent, not a circularity in the derivation. No equation in the paper reduces the output to the input by definition, and no fitted parameter is presented as a prediction. Therefore the derivation chain is self-contained against external benchmarks, and there is no significant circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 3 invented entities

The central claim rests on unpaid premises in two families: (1) the operationalization of 'intent' as frozen-LLM outputs prompted with the authors' own taxonomy, which is neither human-validated nor grounded in an external intent benchmark; and (2) hand-chosen hyperparameters (window size w, fine nodes l, pseudo nodes r, GNN depth, embedding dimension) whose defaults and selection protocol are partly unreported. The invented model components each carry in-paper falsifiable handles through the ablations in Table 3.

free parameters (5)

window size w (sliding-window sentence connectivity) = not stated
Controls Llocal edges in the semantic graph (Section 3.2); hand-chosen hyperparameter whose default value is never reported.
number of fine-grained intent nodes l per coarse node = not stated; swept 2-10 in Figure 3
Controls intent graph expressiveness (Section 3.3); tuned without a stated validation-only protocol.
number of pseudo nodes r = not stated; swept 2-12 in Figure 3
Controls the alignment bottleneck (Section 3.5); tuned without a stated validation-only protocol.
k = 4 coarse intent perspectives (belief, plan, desire, outcome) = 4
Taken from the authors' own intent framework (Wang et al. 2025), not derived from data or an external benchmark.
GNN depth 3, embedding dim 256, max entities 32, learning rate 2e-4, batch size 64 = stated in Section 4.1.3
Standard architecture hyperparameters, reported but without a tuning protocol or sensitivity analysis.

assumptions (5)

domain assumption News intent is a stable, discriminative signal: fake news stems from harmful intent, which is more stable than writing style.
Core premise of the paper, stated in Section 1 ('the inherent purpose of realizing the underlying thoughts, i.e., harmful intent, which in turn provides a stable perspective'). Not measured or validated against any intent benchmark.
domain assumption A frozen LLM prompted with four questions produces valid intent embeddings (C = Generator(t, q)).
Section 3.3. No human evaluation of the generated intents; the paper relies on the prompt quality and the Wang et al. 2025 taxonomy.
ad hoc to paper The belief/plan/desire/outcome taxonomy adequately covers news intent.
Section 4.1.3, self-cited to Wang et al. 2025. The alternative framework ablation (InSide(C')) shows the taxonomy is not unique, weakening but not removing this assumption.
domain assumption SpaCy sentence splitting and entity extraction are accurate enough for graph construction.
Section 3.2; no quality control or error analysis is reported for the extraction step.
domain assumption Dataset labels (PolitiFact, GossipCop, Weibo, LLMFake) are reliable ground truth.
Section 4.1.1; standard practice, but label noise and annotation disagreements are not analyzed.

invented entities (3)

Pseudo nodes Vp and pseudo edges independent evidence
purpose: Bridge semantic and intent graphs via bidirectional attention-based message passing in a common space (Section 3.5).
Falsifiable handle is the w/o DPGA ablation in Table 3, which shows large, consistent performance drops across all four datasets.
Fine-grained intent nodes F (k x l learnable nodes) independent evidence
purpose: Model how coarse intents are implemented in the narrative (Section 3.3).
Falsifiable handle is the w/o Intf ablation in Table 3, which shows consistent drops, and the hyperparameter sweep over l in Figure 3.
Super root node independent evidence
purpose: Global message passing aggregator for both graphs (Section 3.4).
A standard global-node mechanism rather than a new entity, but listed for completeness; the w/o Global ablation in Table 3 provides an in-paper falsifiable handle.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection." pith.science (2026). https://pith.science/paper/R5DJ74B2

@misc{pith2026250901660,
  author       = {Pith},
  title        = {Pith review of: Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R5DJ74B2}},
  note         = {Machine review of arXiv:2509.01660}
}

read the original abstract

Fake news detection is an important and challenging task for defending online information integrity. Existing state-of-the-art approaches typically extract news semantic clues, such as writing patterns that include emotional words, stylistic features, etc. However, detectors tuned solely to such semantic clues can easily fall into surface detection patterns, which can shift rapidly in dynamic environments, leading to limited performance in the evolving news landscape. To address this issue, this paper investigates a novel perspective by incorporating news intent into fake news detection, bridging intents and semantics together. The core insight is that by considering news intents, one can deeply understand the inherent thoughts behind news deception, rather than the surface patterns within words alone. To achieve this goal, we propose Graph-based Intent-Semantic Joint Modeling (InSide) for fake news detection, which models deception clues from both semantic and intent signals via graph-based joint learning. Specifically, InSide reformulates news semantic and intent signals into heterogeneous graph structures, enabling long-range context interaction through entity guidance and capturing both holistic and implementation-level intent via coarse-to-fine intent modeling. To achieve better alignment between semantics and intents, we further develop a dynamic pathway-based graph alignment strategy for effective message passing and aggregation across these signals by establishing a common space. Extensive experiments on four benchmark datasets demonstrate the superiority of the proposed InSide compared to state-of-the-art methods.

Figures

Figures reproduced from arXiv: 2509.01660 by the authors.

**Figure 1.** Comparison between existing works and our proposed framework. Existing detectors focus on surface semantic clues of news and often struggle when the writing pattern shifts. Instead, our method utilizes semantic-intent joint learning, integrating the thoughts (intents) and words (semantics) for enhanced detection. methods typically focus on utilizing such semantic clues to detect fake news. For example, Sheng et al. … view at source ↗

**Figure 2.** Overview of InS ide architecture. Given a news item, InS ide leverages heterogeneous graph structures to represent the news semantics and intent (a-b). After obtaining the encoded semantic and intent representations via dual-level graph updating (c), a dynamic pathway-based graph alignment strategy is utilized to enable cross-signal alignment within a common space (d). The illustrative intent analysis framework is p… view at source ↗

**Figure 3.** Impact of hyperparameters l and r on macF1. Llama-3B Llama-8B Qwen-32B Macro F1 0.7899 0.8352 0.8011 0.8653 0.8220 0.8791 intent w/ semantic (ours) intent w/o semantic ALBERT BERT-base BERT-large Macro F1 0.6090 0.7990 0.7873 0.8653 0.8062 0.8713 semantic w/ intent (ours) semantic w/o intent [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Model performance with and without intent-semantic joint learning across diverse semantic encoders and intent generators, showcasing framework extensibility. 4.4. Hyperparameter Analysis (EQ3) The influence of two key hyperparameters, the number of fine-grained intent …

**Figure 5.** Figure 5: Case visualization analysis of a news item from the Weibo dataset after Graph Alignment. Darker edges indicate critical information pathways. The pseudo nodes (in green) act as dynamic bridges between semantic nodes (in blue) and intent nodes (in orange), enabling flex…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 17 canonical work pages

[6]

Beizhe Hu, Qiang Sheng, Juan Cao, Yuhui Shi, Yang Li, Danding Wang, and Peng Qi

Fake news on Twitter during the 2016 US presidential election.Science, 363(6425):374–378. Beizhe Hu, Qiang Sheng, Juan Cao, Yuhui Shi, Yang Li, Danding Wang, and Peng Qi

work page 2016
[7]

Learning Hierarchical Discourse-level Structure for Fake News Detection. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 3432–

work page 2019
[8]

YichuanLi, KaizeDing, andKyuminLee.2023

ALBERT: A Lite BERT for Self-supervised Learning of Language Representations.CoRR, abs/1909.11942. YichuanLi, KaizeDing, andKyuminLee.2023. GRENADE:Graph-CentricLanguageModelforSelf-Supervised Representation Learning on Text-Attributed Graphs.arXiv preprint arXiv:2310.15109. Bernhard Lutz, Marc Adam, Stefan Feuerriegel, Nicolas Pröllochs, and Dirk Neumann

arXiv 1909
[9]

In Companion Proceedings of The Web Conference 2018, pages 585–593

Detect Rumor and Stance Jointly by Neural Multi-task Learning. In Companion Proceedings of The Web Conference 2018, pages 585–593. International World Wide Web Conferences Steering Committee. 18 Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection Xiaoxiao Ma, Yuchen Zhang, Kaize Ding, Jian Yang, Jia Wu, and Hao Fan

work page 2018
[11]

InProceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 1732–1742

Let silence speak: Enhancing fake news detection with generated comments from large language models. InProceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 1732–1742. QiongNan, DandingWang, YongchunZhu, QiangSheng, YuhuiShi, JuanCao, andJintaoLi.2022. Improving Fake News Detection of Influential Domain via Do...

work page 2022
[14]

InProceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 22–32

DeClarE: Debunking Fake News and False Claims using Evidence-Aware Deep Learning. InProceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 22–32. Qiang Sheng, Juan Cao, Xueyao Zhang, Rundong Li, Danding Wang, and Yongchun Zhu

work page 2018
[15]

InProceedings of the 30th ACM international conference on information & knowledge management, pages 1640–1650

Integrating pattern-and fact-based fake news detection via model preference learning. InProceedings of the 30th ACM international conference on information & knowledge management, pages 1640–1650. Kai Shu, Limeng Cui, Suhang Wang, Dongwon Lee, and Huan Liu. 2019a. defend: Explainable fake news detection. InProceedings of the 25th ACM SIGKDD international ...

arXiv 2004
[17]

Bing Wang, Ximing Li, Changchun Li, Bo Fu, Songwen Pei, and Shengsheng Wang

Qwen3 technical report.arXiv preprint arXiv:2505.09388. Bing Wang, Ximing Li, Changchun Li, Bo Fu, Songwen Pei, and Shengsheng Wang. 2024a. Why Misin- formation is Created? Detecting them by Integrating Intent Features. InProceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 2304–2314. Wei-Yao Wang, Yu-Chieh C...

arXiv

Show all 22 references

[18]

InProceedings of the ACM Web Conference 2024, pages 4128–4137

MSynFD: Multi-hop Syntax aware Fake News Detection. InProceedings of the ACM Web Conference 2024, pages 4128–4137. Bingbing Xie, Xiaoxiao Ma, Jia Wu, Jian Yang, Shan Xue, and Hao Fan

2024
[19]

InProceedings of the ACM Web Conference 2022, pages 2501–2510

Evidence-aware fake news detection with graph neural networks. InProceedings of the ACM Web Conference 2022, pages 2501–2510. Liang Yao, Chengsheng Mao, and Yuan Luo

2022
[20]

InProceedings of the Web Conference 2021, pages 3465–3476

Mining dual emotion for fake news detection. InProceedings of the Web Conference 2021, pages 3465–3476. Yongcheng Zhang, Lingou Kong, Sheng Tian, Hao Fei, Changpeng Xiang, Huan Wang, and Xiaomei Wei

2021
[21]

InProceedings of the 2024 International Conference on Multimedia Retrieval, pages 385–393

Multi-view Counterfactual Contrastive Learning for Fact-checking Fake News Detection. InProceedings of the 2024 International Conference on Multimedia Retrieval, pages 385–393. Zhipu AI

2024
[22]

Accessed: 2025-01-19

GLM-4.https://open.bigmodel.cn/dev/howuse/glm-4. Accessed: 2025-01-19. Xinyi Zhou and Reza Zafarani

2025
[1984]

CanyuChenandKaiShu.2023

Two faces of intention.The Philosophical Review, 93(3):375–405. CanyuChenandKaiShu.2023. CanLLM-GeneratedMisinformationBeDetected? In TheTwelfthInternational Conference on Learning Representations. Chaoqun Cui and Caiyan Jia

2023
[2015]

In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1120–1125

Experiments in open domain deception detection. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1120–1125. Kashyap Popat, Subhabrata Mukherjee, Andrew Yates, and Gerhard Weikum

2015
[2018]

InProceedings of the 2018 World Wide Web Conference, pages 823–831

Me, my echo chamber, and I: introspection on social media polarization. InProceedings of the 2018 World Wide Web Conference, pages 823–831. Haisong Gong, Weizhi Xu, Shu Wu, Qiang Liu, and Liang Wang

2018
[2019]

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. InProceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171–4186...

2019
[2020]

Ling Sun, Yuan Rao, Lianwei Wu, Xiangbo Zhang, Yuqian Lan, and Ambreen Nazir

Credibility-based fake news detection.Disinformation, misinformation, and fake news in social media: Emerging research challenges and Opportunities, pages 163–182. Ling Sun, Yuan Rao, Lianwei Wu, Xiangbo Zhang, Yuqian Lan, and Ambreen Nazir. 2023a. Fighting false information f...

arXiv
[2021]

InProceedings of the AAAI conference on artificial intelligence, volume 35, pages 81–89

Kan: Knowledge-aware attention network for fake news detection. InProceedings of the AAAI conference on artificial intelligence, volume 35, pages 81–89. 17 Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection SongFeng, RitwikBanerjee,...

2012
[2022]

https://openai.com/blog/ chatgpt/

ChatGPT: Optimizing Language Models for Dialogue. https://openai.com/blog/ chatgpt/. Accessed: 2025-01-19. Verónica Pérez-Rosas and Rada Mihalcea

2025
[2024]

InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 508–521

On Fake News Detection with LLM Enhanced Semantics Mining. InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 508–521. Qiong Nan, Juan Cao, Yongchun Zhu, Yanyan Wang, and Jintao Li

2024
[2025]

arXiv preprint arXiv:2501.12948

Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova

arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [6]

Beizhe Hu, Qiang Sheng, Juan Cao, Yuhui Shi, Yang Li, Danding Wang, and Peng Qi

Fake news on Twitter during the 2016 US presidential election.Science, 363(6425):374–378. Beizhe Hu, Qiang Sheng, Juan Cao, Yuhui Shi, Yang Li, Danding Wang, and Peng Qi

work page 2016

[2] [7]

Learning Hierarchical Discourse-level Structure for Fake News Detection. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 3432–

work page 2019

[3] [8]

YichuanLi, KaizeDing, andKyuminLee.2023

ALBERT: A Lite BERT for Self-supervised Learning of Language Representations.CoRR, abs/1909.11942. YichuanLi, KaizeDing, andKyuminLee.2023. GRENADE:Graph-CentricLanguageModelforSelf-Supervised Representation Learning on Text-Attributed Graphs.arXiv preprint arXiv:2310.15109. Bernhard Lutz, Marc Adam, Stefan Feuerriegel, Nicolas Pröllochs, and Dirk Neumann

arXiv 1909

[4] [9]

In Companion Proceedings of The Web Conference 2018, pages 585–593

Detect Rumor and Stance Jointly by Neural Multi-task Learning. In Companion Proceedings of The Web Conference 2018, pages 585–593. International World Wide Web Conferences Steering Committee. 18 Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection Xiaoxiao Ma, Yuchen Zhang, Kaize Ding, Jian Yang, Jia Wu, and Hao Fan

work page 2018

[5] [11]

InProceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 1732–1742

Let silence speak: Enhancing fake news detection with generated comments from large language models. InProceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 1732–1742. QiongNan, DandingWang, YongchunZhu, QiangSheng, YuhuiShi, JuanCao, andJintaoLi.2022. Improving Fake News Detection of Influential Domain via Do...

work page 2022

[6] [14]

InProceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 22–32

DeClarE: Debunking Fake News and False Claims using Evidence-Aware Deep Learning. InProceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 22–32. Qiang Sheng, Juan Cao, Xueyao Zhang, Rundong Li, Danding Wang, and Yongchun Zhu

work page 2018

[7] [15]

InProceedings of the 30th ACM international conference on information & knowledge management, pages 1640–1650

Integrating pattern-and fact-based fake news detection via model preference learning. InProceedings of the 30th ACM international conference on information & knowledge management, pages 1640–1650. Kai Shu, Limeng Cui, Suhang Wang, Dongwon Lee, and Huan Liu. 2019a. defend: Explainable fake news detection. InProceedings of the 25th ACM SIGKDD international ...

arXiv 2004

[8] [17]

Bing Wang, Ximing Li, Changchun Li, Bo Fu, Songwen Pei, and Shengsheng Wang

Qwen3 technical report.arXiv preprint arXiv:2505.09388. Bing Wang, Ximing Li, Changchun Li, Bo Fu, Songwen Pei, and Shengsheng Wang. 2024a. Why Misin- formation is Created? Detecting them by Integrating Intent Features. InProceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 2304–2314. Wei-Yao Wang, Yu-Chieh C...

arXiv