MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641098928 A) filed by Aditya Institute Of Technology And Management; Panduranga Vital Terlapu; Maddula Ratna Mohitha; Palleda Sravani; Murapala Pavitra; Kanumula Mohinidevi; and Ch Suneetha on August 16, 2026, for Retrieval-Augmented Ai System With Dynamic Source Reliability Scoring.
Inventors include Panduranga Vital Terlapu; Maddula Ratna Mohitha; Palleda Sravani; Murapala Pavitra; Kanumula Mohinidevi; and Ch Suneetha.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: The present invention provides a retrieval-augmented AI system (100) configured to determine reliability of retrieved sources. The system (100) comprises a retrieval module (106), source repository (108), source feature extraction module (112), source reliability scoring module (114), source ranking module (130), response generation module (134), feedback/validation module (138), and reliability score updating module (140). The source reliability scoring module (114) determines a reliability score (128) using reliability factors (116) comprising a source consistency factor (118), source freshness factor (120), source authority factor (122), cross- source agreement factor (124), and contextual relevance factor (126). The source ranking module (130) generates a ranked source set (132), and the response generation module (134) generates a response (136). The reliability score updating module (140) updates the dynamic reliability score (128) based on validation information.
Disclaimer: Curated by HT Syndication.
