AUT Journal of Modeling and Simulation

AUT Journal of Modeling and Simulation

Neuro-Symbolic Claim Detection for Arabic: Integrating Graph Neural Networks with Efficiently Fine-Tuned Small Language Models

Document Type : Research Article

Authors
Dina Falah Masood 1
Heshaam Faili 2
1 Department of Computer Engineering, College of Alborz, University of Tehran, Tehran, Iran.
2 School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran
10.22060/miscj.2025.24744.5431
Abstract
The rise in online information has increased the spread of misinformation and disinformation, which makes claim detection important for the reliability of digital content. Claim detection refers to the problem of determining whether a factual statement is real or fake. While there has been great advancement in high-resource languages, Arabic is particularly challenging due to its special characteristics and the low availability of annotated resources. In this paper, we make use of the Arabic News Stance (ANS) corpus and develop a multi-stage framework to help tackle these problems. Our method effectively models the structural interplay between claims and words with a carefully heterogeneous bipartite graph and the use of advanced Graph Neural Networks (GNNs). In parallel, we used two Small-scale Language Models (SLMs) in this domain, namely LLaMA-3.2-1B and Qwen2.5-3B, and demonstrated that together with QLoRA-based fine-tuning, these lightweight models can be as effective as or even better than heavier baselines. To this point, we suggested a novel neuro-symbolic fusion paradigm that unifies structural reasoning with linguistic prior information. By integrating the outputs from a Graph Convolutional Network (GCN) with the fine-tuned Qwen2.5-3B logits, we achieved the state-of-the-art performance on the ANS dataset with an accuracy of 0.732 and an F1-score of 0.717. This framework shows that the combined results of structural graph reasoning and SLMs lead to accurate and computationally efficient claim detection. This work not only enhances the detection of Arabic misinformation but also provides a scalable solution for low-resource languages.
Keywords
Claim detection
Arabic
Graph Neural Networks
Small-scale Language Models
Supervised fine-tuning
Subjects
[1] E.C. Choi, E. Ferrara, Fact-gpt: Fact-checking augmentation via claim matching with LLMs, in:  Companion Proceedings of the ACM Web Conference 2024, 2024, pp. 883-886.
[2] W.-Y. Kao, A.-Z. Yen, How we refute claims: Automatic fact-checking through flaw identification and explanation, in:  Companion Proceedings of the ACM Web Conference 2024, 2024, pp. 758-761.
[3] A. Gupta, A. Singhal, T. Law, V. Rao, E. Duan, R.L. Li, Can LLMs Verify Arabic Claims? Evaluating the Arabic Fact-Checking Abilities of Multilingual LLMs, in:  Proceedings of the 1st Workshop on NLP for Languages Using Arabic Script, 2025, pp. 104-113.
[4] A. Farghaly, The Arabic language, Arabic linguistics and Arabic computational linguistics, Arabic computational linguistics,  (2010) 43-81.
[5] A. Alshehri, A. AlShabeb, Exploring attitudes, identity, and linguistic variation among Arabic speakers: Insights from acoustic landscapes, International Journal of Arabic-English Studies, 24(2) (2023) 1-16.
[6] N.A. Othman, D.S. Elzanfaly, M.M.M. Elhawary, Arabic Fake News Detection Using Deep Learning, IEEE Access, 12 (2024) 122363-122376.
[7] L. Majer, J. Šnajder, Claim Check-Worthiness Detection: How Well Do LLMs Grasp Annotation Guidelines?, in:  Proceedings of the Seventh Fact Extraction and Verification Workshop (FEVER), 2024, pp. 245-263.
[8] Z.M. Hüsünbeyi, T. Scheffler, Ontology Enhanced Claim Detection, CoRR,  (2024).
[9] T. Chen, A. Stefanidis, Z. Jiang, J. Su, Improving biomedical claim detection using prompt learning approaches, in:  2023 IEEE 4th International Conference on Pattern Recognition and Machine Learning (PRML), IEEE, 2023, pp. 369-376.
[10] L.-H. Lee, Y.-H. Cheng, J.-H. Yang, K.-Y. Tien, NCUEE-NLP at SemEval-2023 Task 8: Identifying Medical Causal Claims and Extracting PIO Frames Using the Transformer Models, in:  Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), 2023, pp. 312-317.
[11] A. Wührl, R. Klinger, Entity-based Claim Representation Improves Fact-Checking of Medical Content in Tweets, in:  Proceedings of the 9th Workshop on Argument Mining, 2022, pp. 187-198.
[12] A. Wührl, R. Klinger, Claim Detection in Biomedical Twitter Posts, in:  Proceedings of the 20th Workshop on Biomedical Language Processing, 2021, pp. 131-142.
[13] J. Khouja, Stance Prediction and Claim Verification: An Arabic Perspective, in:  Proceedings of the Third Workshop on Fact Extraction and VERification (FEVER), 2020, pp. 8-17.
[14] T. Zerrouki, pyarabic, An Arabic language library for Python,  (2010).
[15] Llama-3.2-1B-Instruct,  (Sept 25, 2024).
[16] Q. Team, Qwen2.5: A Party of Foundation Models,  (2024).
[17] S.E. Sorour, H.E. Abdelkader, AFND: Arabic fake news detection with an ensemble deep CNN-LSTM model, Journal of Theoretical and Applied Information Technology, 100(14) (2022) 5072-5086.
[18] L. Yao, C. Mao, Y. Luo, Graph convolutional networks for text classification, in:  Proceedings of the AAAI conference on Artificial Intelligence, 2019, pp. 7370-7377.
[19] A.G. Vrahatis, K. Lazaros, S. Kotsiantis, Graph attention networks: a comprehensive review of methods and applications, Future Internet, 16(9) (2024) 318.
[20] W. Hamilton, Z. Ying, J. Leskovec, Inductive representation learning on large graphs, Advances in neural information processing systems, 30 (2017).
[21] J. Du, S. Zhang, G. Wu, J.M. Moura, S. Kar, Topology Adaptive Graph Convolutional Networks,  (2018).
[22] K. Xu, W. Hu, J. Leskovec, S. Jegelka, How Powerful are Graph Neural Networks?, in:  International Conference on Learning Representations, (2018).
[23] J. Gasteiger, A. Bojchevski, S. Günnemann, Predict then Propagate: Graph Neural Networks meet Personalized PageRank, in:  International Conference on Learning Representations (2018).
AUT Journal of Modeling and Simulation
Volume 58, Issue 1
2026
Pages 23-32