MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202621080942 A) filed by Symbiosis International Deemed University on June 30, 2026, for Multilingual Multimodal Media Summarization System Using Gnn And Transformer Models.
Inventors include Avantika R. Patil; Pratik P. Gunjal; Nishit Bohra M; Dr. Aniket K. Shahade; Dr. Priyanka V. Deshmukh; Dr. Shruti Patil; and Dr. Ketan Kotecha.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: ABSTRACT MULTILINGUAL MULTIMODAL MEDIA SUMMARIZATION SYSTEM USING GNN AND TRANSFORMER MODELS The present invention discloses a multilingual multimodal media summarization system (100) for summarizing video and audio content using GNN-enhanced extraction and transformer- based abstraction. The system (100) comprises an input processing module (110) with Whisper-based transcript extraction, a preprocessing module (120) with multimodal feature extraction including speaker diarization, audio prosody, and visual keyframes, an extractive summarization module (130) with TF-IDF extractive unit (132) and GNN-based extractive unit (134) using graph attention networks with salience scoring unit (136), an abstractive summarization module (140) with mT5 transformer unit (142) and hybrid summarization unit (144), an evaluation module (160) with comprehensive metrics and statistical testing, and an output dashboard module (170). The system achieves ROUGE-L F1 up to 0.5345 and BERTScore F1 exceeding 0.81 across 150 multilingual videos. [
Disclaimer: Curated by HT Syndication.
