MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641101348 A) filed by G Pullaiah College Of Engineering And Technology on August 21, 2026, for A Novel Multimodal Data Fusion For Real-Time Sentiment Analysis In Social Media.
Inventors include M Jayasunitha; D K Shareef; V. Sravani; K Lakshmi; P. Rajya Lakshmi; K. Manjusha; V. Raja Sekhar; and G Lokesh.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: A novel multimodal data fusion framework is disclosed for performing real time sentiment analysis on social media content. The system continuously ingests heterogeneous data streams comprising textual messages, image and video frames, and interaction metadata. Separate modality specific encoders extract semantic, visual affective, and behavioral features which are subsequently aligned in a shared latent space through a dynamic cross modal attention fusion module. A lightweight temporal aggregator maintains short term context while a decision head produces continuous sentiment scores together with polarity and intensity labels. Adaptive resource allocation and incremental model updating mechanisms ensure operation under streaming constraints with bounded latency. Experimental evaluation on live platform derived streams demonstrates improved accuracy and responsiveness relative to unimodal or static fusion baselines while preserving privacy conscious processing. The framework supports deployment in monitoring, early warning, and interactive analytics scenarios that demand timely multimodal understanding of social discourse
Disclaimer: Curated by HT Syndication.
