Jul 2026· 2026 6th International Conference on Inventive Computation and Information Technologies (ICICIT)· pp. 1421-1426· 0 citations· 17 references
Abstract
The rapid proliferation of digitally distributed news has made large-scale automated verification an urgent research priority, as deceptive content spreads across social platforms faster than manual reviewers can evaluate it. This work introduces a two-stage deep learning pipeline in which a pretrained BERT encoder generates rich contextual token embeddings that are subsequently processed by a Bidirectional Long Short-Term Memory (BiLSTM) network, enabling the classifier to exploit both sentence-level semantics and document-level narrative flow simultaneously. Experiments on the WELFake benchmark [1] following a standardised cleaning pipeline—lowercasing, URL stripping, tokenisation, stop-word removal, and WordNet lemmatisation— yielded 98.7 % accuracy and an F1-score of 0.986. This surpasses a BERT-only baseline by 3.1 percentage points and exceeds several previously published state-of-the-art results. The gains confirm that coupling transformer-based semantic representations with recurrent sequential modelling produces a measurable and reproducible improvement in misinformation detection. Planned extensions include multilingual evaluation and knowledge-distilled encoder replacements targeting real-time throughput constraints.
The rapid spread of misinformation across digital platforms has made fake news detection a critical challenge, as it
can influence public opinion, disrupt social stability, and reduce trust in credible information sources. Manual verification is no
longer feasible at scale due to the large volume of content generated daily. Existing approaches have explored hybrid
architectures combining transformer-based models such as Bidirectional Encoder Representations from Transformers (BERT)
with sequential models like Long Short-Term Memory (LSTM) for fake news classification; however, such approaches may have
limitations in capturing the diverse linguistic, contextual, and structural patterns present in textual data. To address this
limitation, this paper proposes a hybrid transformer-based ensemble model for automated fake news identification using the
FakeNewsNet dataset. The proposed system integrates BERT with LSTM for contextual and sequential learning, Robustly
Optimized BERT Pretraining Approach (RoBERTa) for improved textual representation, and Light Gradient Boosting Machine
(LightGBM) for learning statistical patterns from textual features. Individual models provide strong baseline performance, while
the ensemble combines their predictions using a weighted strategy to improve overall accuracy and robustness. Experimental
results show that the ensemble model achieves an accuracy of approximately 93%, outperforming the individual constituent
models. The system can be applied in real-time news verification platforms to assist users, journalists, and fact-checkers in
identifying misleading information more effectively
M. E. Babu, G. Sukanya· International Journal for Re...· 0 citations
While existing ensemble-based fake news detection models achieve high accuracy on benchmark datasets, they suffer from critical linguistic and preprocessing bottlenecks. The reliance on case-sensitive regex for entity extraction causes failure on lowercased or vernacular (Hinglish) inputs. Furthermore, hardcoded entity+year fallback mechanisms fail for hyphenated scientific missions (e.g., Aditya-L1) and future-dated events (e.g., 2025 summits) due to case mismatches and missing temporal markers in Wikipedia snippets. This paper proposes a robust preprocessing pipeline incorporating Google Translation, spaCy-based NER with hyphenated-word normalization, and a future-event-aware fallback logic. By replacing the heavy BART-large-MNLI with a lightweight DeBERTa-v3-base cross-encoder, we achieve a 40 percent reduction in inference latency. Experimental results demonstrate that the proposed system retains 99.68 percent accuracy on the ISOT benchmark while achieving 100 percent classification on a diverse Hinglish/low-resource test suite—closing the linguistic generalization gap left by previous works.
Rishabh Kumar, Aditya Kumar· Journal of Frontiers in Mult...· 0 citations
The rapid growth of online media has accelerated the circulation of misinformation, posing serious risks to public trust, social stability, and democratic processes. Detecting fake news remains a challenging task, particularly because traditional machine learning approaches such as Support Vector Machines and Naïve Bayes rely on shallow linguistic cues and often fail to recognize sarcasm, implicit tone, or evolving writing styles. Earlier deep learning models, including CNNs and LSTMs, provide modest improvements but continue to struggle with long text and shifting topical contexts. This study examines the effectiveness of two transformer-based architectures—BERT and DeBERTa—for identifying fake news using only textual information from headlines and article bodies. To enhance generalization, news articles from ten publicly available datasets are consolidated into a single corpus namely FakeDiverse, that exposes the models to a broader spectrum of linguistic patterns. MinHash similarity along with Locality Sensitive Hashing algorithm has been employed to evaluate the lexical redundancy within the FakeDiverse corpus as a post hoc analysis. Class imbalance between real and fake news is addressed through a weighted cross-entropy loss function, ensuring that minority-class instances are adequately represented during training. FakeDiverse dataset is split in the ratio of 80:20 for training and testing, and subsequently the models BERT and DeBERTA are trained on the training set and evaluated using the test sets. Experimental results demonstrate that both the models capture contextual nuances effectively, with BERT achieving 98% accuracy and DeBERTa reaching 99%. The findings demonstrate that BERT and DeBERTa achieves strong performance on FakeDiverse corpus, and cross-dataset evaluation shows that models have challenges to generalize unseen data distribution thereby highlighting the need for enhanced generalization strategies as well as domain adaptation.
Archana Praveen Kumar, A. S, Akshara G. Bhat et al.· Scientific Reports· 0 citations
The proliferation of social media has created fertile ground for misinformation, a challenge further intensified by recent advances in generative artificial intelligence. Modern fake news increasingly takes the form of sophisticated multimodal campaigns, where synthetic images and stylistically manipulated text are jointly employed to evade existing detection systems. Despite substantial progress, real-world deployment of multimodal fake news detection models remains constrained by an ''impossible triangle'' of accuracy, inference efficiency, and robustness. To address these challenges, we propose DAR-Lite, a serial two-stage framework that rethinks the detection pipeline through explicit decoupling of representation denoising and contextual reasoning. In the first stage, a variational information bottleneck distills compact and noise-invariant semantic anchors from raw image--text inputs, reducing sensitivity to nuisance factors. In the second stage, an adaptive reasoning engine integrates retrieval-augmented verification, social credibility propagation, and dynamic propagation signals via gated cross-attention to perform structured reasoning over heterogeneous contexts. An auxiliary logical fallacy detection task further encourages reasoning beyond surface-level pattern matching. Extensive experiments on multiple large-scale benchmarks demonstrate that DAR-Lite consistently outperforms state-of-the-art methods, particularly under low false-positive constraints critical for real-world applications. Efficiency analyses further show that the proposed serial architecture achieves a favorable balance between detection performance and computational cost, making DAR-Lite suitable for practical, large-scale misinformation detection.
Maolin Wang, Ziting Mai, Zichun Liu et al.· Proceedings of the 32nd ACM...· 0 citations
The rapid proliferation of fake news across social media and messaging platforms poses a serious threat to information
integrity, public discourse, and institutional trust. Automated detection research has progressed through linguistic feature-based
methods, recurrent and attention-based neural architectures, word-embedding strategies, and, increasingly, hybrid systems that
fuse multiple complementary signals. This paper presents a comprehensive review of 30 studies spanning foundational
linguistic-cue research, classical machine learning, RNN/LSTM/Bi-LSTM architectures, transformer-augmented models,
propagation- and stance-based methods, adversarial robustness, and explainability-oriented approaches. We organize these
works into a structured taxonomy, compare them across accuracy, interpretability, computational cost, and real-time suitability,
and synthesize ten recurring research gaps: limited real-time readiness, poor explainability, weak performance on short informal
text, fragmented multi-signal integration, vulnerability to sophisticated fake content, high computational cost, weak crossdomain generalization, an unresolved accuracy/efficiency/interpretability trade-off, the absence of a principled safeguard
against ensemble override of factual contradictions, and lack of resilience to external verification-service failure. Building on
this synthesis, we formulate a precise problem statement and propose a hybrid multi-signal methodology that integrates heuristic
linguistic analysis, Bi-LSTM-based contextual modelling, real-time factual verification with deterministic offline fallback, and a
decision safeguard mechanism (Veto Logic) within an explainable decision framework. A mathematical formulation including
the override condition, a fusion model, and an algorithmic procedure for the proposed framework are presented, providing the
complete conceptual and methodological foundation for an experimentally validated hybrid detection system — TruthLens —
reported in our companion result paper.
Zafeer Ahmed Patel, Ashwini S. Gaikwad· International Journal for Re...· 0 citations
The online media has proliferated and become more accessible, so too has the ease with which misinformation can spread and be consumed and automated mechanisms to detect this form of online deception will be very important research targets moving forward. The implementation of new techniques from deep learning and transformer models pre-trained on large amounts of data has greatly improved the ability to detect misinformation, however many detectors are hindered by limitations on their ability to utilize semantic features and interpret the resulting predictions. In this study, we present our Adaptive Semantic Feature Refinement for Explainable Fake News Detection Utilizing Pre-Trained Transformers model, which uses a novel hybrid deep learning architecture that combines DeBERTa-v3, Bidirectional Gated Recurrent Unit (BiGRU), Multi-Head Self-Attention, and an Adaptive Semantic Refinement Module (ASRM) to produce higher-quality representations of the text used to classify fake news articles into binary categories. In addition, by applying SHAP (SHapley Additive exPlanations) values and Integrated Gradients to improve prediction transparency, we were able to produce both global and local explanations of the model's predictions. Our model was tested using a large corpus of fake news articles that included 682,661 articles, the resulting accuracy was 85.85%, with a ROC AUC statistic of .9294. Overall, results indicate that our proposed architecture successfully combines the process of refining the semantic features of text data while providing an explainable artificial intelligence solution for real-world applications of fake news detection.
Sudha Patel, Shilpa Serasiya, Sachi Bhavsar et al.· International journal of com...· 0 citations