MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202621094605 A) filed by Symbiosis International Deemed University on August 04, 2026, for Trustworthy Retrieval-Augmented Generation System For Semantic Document Search.
Inventors include Dr. Sagarkumar Badhiye; Dr. Pradnya Borkar; Ananya Prasad; Shriya Yerane; and Omkar Kamble.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: ABSTRACT TRUSTWORTHY RETRIEVAL-AUGMENTED GENERATION SYSTEM FOR SEMANTIC DOCUMENT SEARCH The present invention discloses a trustworthy retrieval-augmented generation system (100) for semantic document search and information retrieval from PDF documents. The system (100) comprises a user interface module (110), a document upload module (120), a text extraction module (130), a text preprocessing module (140) configured to perform text cleaning, tokenization, overlapping chunking with context preservation and normalization, an embedding generation module (150) employing a transformer-based sentence embedding model, a vector indexing and storage module (160) utilizing a FAISS index, a query processing module (170), a semantic search and retrieval module (180) and a results display module (190). The system is characterized in that it employs overlapping text chunking combined with lightweight transformer-based sentence embeddings and FAISS vector indexing to achieve trustworthy, transparent and semantically accurate document retrieval wherein retrieved text segments are directly traceable to the source document content, eliminating the risk of hallucinated information. [
Disclaimer: Curated by HT Syndication.
