MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641076184 A) filed by Sr University on June 19, 2026, for A System For Efficient Sarcasm And Irony Detection On Social Media Platforms.
Inventors include Sravan Kumar Devulapalli; and Suresh Kumar Mandala.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: A SYSTEM FOR EFFICIENT SARCASM AND IRONY DETECTION ON SOCIAL MEDIA PLATFORMS An automated computational system and method are disclosed for detecting sarcasm and irony in social media text streams via a lightweight hybrid architecture. The system processes raw text inputs through a preprocessing layer before routing data into parallel contextual and temporal feature extraction paths. A deep context path utilizes a Bidirectional Encoder Representations from Transformers (BERT) model for bidirectional encoding, while a temporal path uses integrated Recurrent Neural Network (RNN) and Long Short- Term Memory (LSTM) layers to extract sequential learning and long-term text patterns. The outputs are combined in a feature fusion layer and subsequently compressed via an optimization layer using quantization and pruning mechanisms. This framework yields an exceptionally compact model tailored for low-power, real- time deployment on mobile and edge devices, classifying text streams into sarcasm, irony, or neutral designations.
Disclaimer: Curated by HT Syndication.
