MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202521007789 A) filed by Tata Consultancy Services Limited on January 30, 2025, for Statistical And Natural Language Techniques Based Generation Of Optimized Prompts.
Inventors include Panigrahy, Sanjaya Kumar; Pattanaik, Bhabani Sankar; and Pradhan, Soubhagya Ranjan.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT STATISTICAL AND NATURAL LANGUAGE TECHNIQUES BASED GENERATION OF OPTIMIZED PROMPTS Prompt engineering is one of the critical steps in Generative artificial intelligence (GenAI) where a prompt engineer writes and evaluates prompts in an iterative manner to find the best prompt for a given use case. Conventional approaches fail to construct required prompts with minimal input from users. Optimization of the prompts is performed by hit and trial method using prompt engineering skills. Present disclosure provides systems and methods that provide options for appropriate blocks selection and generation of blocks. The generated blocks are mapped to prompt templates. Individual blocks are optimized to generate candidate optimized prompts using statistical and natural language processing techniques until all templates are processed. All generated candidate optimized prompts are evaluated against selected Large Language Model(s) (LLMs) and one or more identified/selected evaluation metrics are calculated for each output. Prompt with best metrics is chosen as final optimized prompt for subsequent processing and analysis. [
Disclaimer: Curated by HT Syndication.
