MUMBAI, India, April 17 -- Intellectual Property India has published a patent application (202641043291 A) filed by Marri Laxman Reddy Institute Of Technology And Management, Hyderabad, Telangana, on April 4, for 'a hardware-integrated adaptive multi-modal knowledge extraction system using hybrid retrieval and local large language model orchestration.'
Inventor(s) include Dr M Nagalakshmi; Sriramoju Krishna; Nalajala Prabhas; and Yendluri Venkat Karthik.
The application for the patent was published on April 17, under issue no. 16/2026.
According to the abstract released by the Intellectual Property India: "The present invention discloses a hardware-integrated adaptive multi-modal knowledge extraction system configured to process heterogeneous data sources including URLs, documents, and textual inputs. The system comprises a hardware-accelerated preprocessing engine, a hybrid retrieval engine combining statistical and semantic techniques, an adaptive context selection module, and a local large language model (LLM) orchestration engine. The system performs intelligent segmentation, vectorization, and context-aware retrieval to generate optimized prompts for local LLM inference. The invention enables real-time response generation with reduced latency, enhanced contextual accuracy, and improved data privacy through fully offline processing. The system further supports dynamic model selection, memory-optimized storage, and scalable architecture for efficient knowledge extraction applications."
Disclaimer: Curated by HT Syndication.
