2026· ITEGAM- Journal of Engineering and Technology for Industrial Applications (ITEGAM-JETIA)· Vol 12, pp. 1468-1475· 0 citations
TL;DR
A robust and scalable framework for real-time rumor detection powered by Large Language Models, like BERT, RoBERTa, and GPTs-4, which combines Natural Language Processing techniques with sentiment analysis, stance detection, and automated fact-checking to enhance contextual understanding and assess credibility more effectively.
Abstract
The rapid increase in the content generated by users on the social media has significantly accelerated the dissemination of rumors and misinformation, creating serious societal challenges. This paper introduces a robust and scalable framework for real-time rumor detection powered by Large Language Models, like BERT, RoBERTa, and GPTs-4. The proposed system combines Natural Language Processing techniques with sentiment analysis, stance detection, and automated fact-checking to enhance contextual understanding and assess credibility more effectively. Data is collected from various social media platforms like Twitter, Facebook, and Reddit, along with benchmark datasets like PHEME and FakeNewsNet. Experimental results show that LLMs significantly outperform traditional machine learning models, with GPT-4 achieving an accuracy of up to 94.7%. An ablation study further highlights the contribution of each module to the overall performance of GPT -4 for Rumor Detection. This framework offers a very comprehensive and flexible approach for countering misinformation across a range of languages and social environments.
The majority of social media research relies on datasets gathered from social media websites such as Reddit and Twitter. However, their intrinsic high noisy content reduces the performance of data-driven models. This restriction of noisy data has made data preparation techniques necessary. Current systems usually need to pay more attention to data quality and their performance heavily depends on noisy social media datasets. In this work, we concentrate on creating high-quality social media data for rumor detection tasks on the widely popular PHEME-9 dataset. Eliminating noisy responses is essential for misinformation tasks since it guarantees that the model has been trained on precise and pertinent data, improving its capacity to identify and validate rumors successfully. Large language models (LLMs) are used in this work to filter out irrelevant comments prior to the application of machine and deep learning techniques. This bifurcated approach helps to improve model accuracy and lower computing burden. We assume that this will further help in accurate rumor identification and can support environmental sustainability using less computational resources. Our proposed methodology shows an average improvement in the trained filtered models’ performance in terms of accuracy and F1-scores on six events in the PHEME-9 dataset. Further, to validate the effectiveness of the trained model, we performed interpretability and error analysis.
Shakshi Sharma, Anjali Goyal, Naman Ahuja et al.· International Journal of Com...· 0 citations
Social media has become a significant part of people’s lives, leading to the rapid spread of false information, such as rumors, which negatively impact society and individuals. Therefore, it is crucial to detect such rumors at an early stage. This study proposes a novel approach for explainable rumor detection by integrating topic modeling with Local Interpretable Model-agnostic Explanations (LIME). Our approach employs an unsupervised machine learning technique, specifically Latent Dirichlet Allocation (LDA), to uncover hidden topics within rumor data. These topics serve as features for classification. To ensure stability, we utilize a Random Forest classifier with 5-fold cross-validation, achieving a superior accuracy of 93.25% on the PHEME dataset compared to other state-of-the-art models. The combination of topic-based classification enhances the accuracy and interpretability of rumor detection model. Additionally, our model offers greater interpretability than traditional LIME methods. While LIME provides local explanations that may vary for each instance, our method captures stable topic distributions that are particularly effective for early-stage rumor detection with better explanations.
Barsha Pattanaik, Sourav Mandal, R. M. Tripathy et al.· Discover Computing· 0 citations
— With the widespread adoption of social media, the speed and reach of information dissemination have expanded dramatically in unpredictable ways. This is not merely a simple phenomenon; it is fundamentally changing the way we acquire and process information. Stance detection in social media text is therefore particularly important, especially in rapidly changing and emotionally charged online discussions. Stance detection goes beyond mere sentiment analysis; it involves profoundly analyzing the complex attitudes, stances, and implicit biases embedded in the text. In recent years, with the rapid development of Large Language Models (LLM) and deep learning technologies, traditional stance detection methods have gradually been replaced by more complex and sophisticated techniques, particularly in large-scale text processing, where deep learning plays an increasingly prominent role. This paper will explore the construction of stance data, the evolution of model paradigms, and the issue of generalization in real-world applications. The challenge of stance detection lies not only in accurately identifying stances but also in integrating various factors — such as emotion, culture, and social context — into the model.This paper will analyze this from multiple levels, including data construction and technological changes, and propose new theoretical paths to improve the practicality and credibility of this field.
Jiayi Zhao· International journal of eng...· 0 citations
Misinformation on social media can be a severe threat to social trust, safety, and health of the population, especially in times of an epidemic like the Monkeypox outbreak. This study specifically focuses on a hybrid RoBERTa–GRU architecture designed to capture both contextual semantics and temporal dependencies in social media discourse. This research presents how the combination of text mining and social network analysis enables Artificial Intelligence (AI) to support misinformation detection. The proposal of a hybrid architecture that integrates RoBERTa and GRU-based embeddings in a contextual fashion and GRU-based modelling of sequential patterns helps identify and substantiate misinformation in social media posts. Based on a curated X (formerly Twitter) dataset consisting of Monkeypox posts (5787 posts), the model provided state-of-the-art results, with ROC-AUC 0.9979 and Cohen’s kappa 0.9887; standalone baselines were also surpassed. Results reveal that the proposed transformer–RNN hybrid effectively captures both semantic depth and temporal relationships in misinformation detection tasks. In addition to performance, the paper addresses the limitations of dataset bias, multilingual constraint issues, and scalability as related to cross-linguistic applicability, multimodal study, and performance in real-time and resource-limited systems. The study adds value in this emerging body of knowledge on AI-driven social media analytics by offering practical guidance for mitigating health-related misinformation online.
Arafat Rohan, Md Asraful Islam, Areyfin Mohammed Yoshi et al.· Scientific Reports· 0 citations
ABSTRACT - Rumours and misinformation propagate rapidly across online social networks, posing significant challenges to maintaining the integrity of information dissemination. In recent years, machine learning (ML) techniques have emerged as promising tools for automating the detection and mitigation of rumours. This review paper provides a comprehensive examination of the advancements in rumour detection using ML approaches. The paper begins by outlining the landscape of rumour dissemination in online social networks, highlighting the characteristics and challenges associated with rumour detection. Subsequently, it systematically categorizes and analyzes various ML methods employed for rumour detection, including supervised, unsupervised, and semi-supervised learning approaches. Furthermore, the review delves into the diverse features and representations utilized in ML models for rumour detection, such as textual content, user engagement patterns, network structures, and temporal dynamics. It discusses the strengths and limitations of different feature sets and their impact on the effectiveness of rumour detection systems. Moreover, the paper explores the intricacies of dataset construction and evaluation methodologies for training and testing rumour detection models. It examines commonly used benchmark datasets and evaluation metrics, emphasizing the importance of robust evaluation frameworks for assessing the performance of ML-based rumour detection systems accurately. Additionally, the review identifies key challenges and open research questions in the field of rumour detection using ML, including handling evolving rumour patterns, addressing adversarial attacks, and enhancing the interpretability and explain ability of ML models. It also discusses potential directions for future research aimed at advancing the state-of-the-art in rumour detection and mitigation.
C.Sandeep Reddy, Dr.Kiran B.M, P. Rani· International Scientific Jou...· 0 citations
A multi-model learning framework that combines the complementary strengths of classical machine learning classifiers, deep sequential neural networks, and transformer-based contextual language models to detect fake news on social media is proposed.
Priya Verma· International Journal of Res...· 0 citations