MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641009366 A) filed by Sr University, Warangal, Telangana, on Jan. 29, for 'a cross-domain deep twin learning-based system and method for personalized social recommendations using user opinion data.'
Inventor(s) include Dr. Lakshmikanth Paleti; and Dr. Balajee Maram.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to an intelligent cross-domain recommendation system designed to address critical limitations in existing personalized social recommendation technologies arising from the exponential growth of user-generated opinion data across multiple digital platforms. With the widespread use of social media, e-commerce portals, online forums, and content-sharing services, billions of users generate massive volumes of reviews, ratings, comments, and interaction data daily, making accurate personalization increasingly challenging. Globally, more than 4.9 billion users contribute over 500 million opinion-based interactions per day, yet studies indicate that nearly 70% of users are dissatisfied with the relevance of recommendations and approximately 60% abandon platforms due to ineffective personalization. Existing recommendation systems predominantly rely on single-domain collaborative filtering, content-based filtering, or shallow machine learning techniques, which are limited in capturing complex user preferences, suffer from data sparsity and cold-start problems, and fail to utilize cross-domain correlations effectively. These limitations result in poor recommendation accuracy, biased outcomes, and reduced scalability. To overcome these drawbacks, the present invention proposes a cross-domain deep twin learning-based system and method that simultaneously learns shared and domain-specific user preference representations from heterogeneous opinion data. The twin learning architecture enables effective knowledge transfer across domains, allowing the system to leverage shared behavioural patterns while preserving domain-specific characteristics. By integrating advanced deep learning models and a cross-domain knowledge transfer mechanism, the proposed invention significantly reduces sparsity and cold-start effects, adapts to evolving user behaviour, and improves personalization accuracy. The invention thereby provides a scalable, adaptive, and reliable solution for generating highly relevant personalized social recommendations across diverse digital environments."
Disclaimer: Curated by HT Syndication.
