MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202631059278 A) filed by Purbani Kar; and Dr. Swapan Debbarma on May 09, 2026, for Hybrid Optimization And Reinforcement Learning Framework For Cross-Lingual Hate Speech Detection.
Inventors include Purbani Kar; and Dr. Swapan Debbarma.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: Hybrid Optimization and Reinforcement Learning Framework for CrossLingual Hate Speech Detection This invention describes a hybrid computational framework for detecting hate speech across multiple languages in social media environments. The system addresses challenges such as multilingual variability, limited labeled data, and contextual ambiguity in hate speech detection. The proposed approach integrates an Improved Lion Swarm Optimization (ILSO) algorithm with a deep residual network architecture for extracting language-independent deep textual features. Feature dimensionality is further refined using a Dynamic Sparrow Search (DSS) algorithm to select highly informative attributes while reducing redundancy. For classification and decision-making, an Ensemble Deep Reinforcement Learning (EDRL) mechanism aggregates predictions from multiple learning models to enhance detection accuracy and minimize misclassification. The system is designed to operate across diverse languages and cultural contexts by capturing shared semantic representations of hate speech. Experimental evaluations demonstrate improved accuracy, precision, and recall in cross-lingual hate speech detection tasks, highlighting the effectiveness of the hybrid optimization and reinforcement learning strategy for multilingual social media moderation.
Disclaimer: Curated by HT Syndication.
