MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641101727 A) filed by Dr. Ramesh Shahabadkar on August 23, 2026, for Nlp Engine For Regional And Low-Resource Languages.
Inventors include Dr. Ramesh Shahabadkar; Pallavi K V; Dr. Niranjan Kumar; Veena K; and Bhavya Balakrishnan.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: The present invention relates to an intelligent Natural Language Processing (NLP) engine for regional and low-resource languages, enabling language understanding, generation, translation, and speech-based interaction. The engine integrates transformer-based language models, adaptive transfer learning, multilingual embeddings, and rule-based linguistic processing to address limited annotated linguistic resources. An automated data augmentation framework generates synthetic corpora using contextual paraphrasing, back translation, and phonetic normalization, thereby improving model robustness and reducing dependence on manually labelled datasets. A dynamic language adaptation module updates lexical, semantic, and syntactic representations using user interactions while preserving privacy through federated learning. The engine further provides speech-to-text conversion, text-to-speech synthesis, sentiment analysis, named entity recognition, document summarization, question answering, and multilingual translation optimized for regional dialects. The engine is deployable across cloud, edge, and mobile platforms with reduced computational requirements for diverse applications.
Disclaimer: Curated by HT Syndication.
