MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202621049643 A) filed by Javeriya Naaz Ishtiyaque Syed; Dr. Ranjit Ramesh Keole; Mr. Swapnil Shriramji Nehar; and Mr. Rahul Rambhau Bhoge on April 18, 2026, for A System And Method For Machine Learning-Based Classification Of Multimodal Fact-Checked Misinformation On Social Networks.
Inventors include Javeriya Naaz Ishtiyaque Syed; Dr. Ranjit Ramesh Keole; Mr. Swapnil Shriramji Nehar; and Mr. Rahul Rambhau Bhoge.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: The present invention relates to a computer-implemented system and method for classification of multimodal fact-checked misinformation on social networks. The system receives content records including claim text, evidence text, social-network posts, and metadata. The records are preprocessed by removing incomplete entries, consolidating text fields, encoding labels, and reducing textual noise. Textual features are extracted using TF-IDF, and structured meta-features including claim length, evidence count, and post count are generated. The features are combined and processed using an ensemble classifier comprising Random Forest, Gradient Boosting, and a stacking architecture with a meta-learner. The system outputs a misinformation class label for automated content assessment. The invention provides an efficient technical framework for improved classification of noisy and heterogeneous multimodal social-network content.
Disclaimer: Curated by HT Syndication.
