MUMBAI, India, Sept. 12 -- Intellectual Property India has published a patent application (202421015859 A) filed by Tata Consultancy Services Limited, Maharashtra, on March 6, 2024, for 'transpiling source code to functionally correct and readable in target language using large language models.'
Inventor(s) include Gandhi, Shubham Rajnish; Patwardhan, Manasi Samarth; Khatri, Jyotsana; Vig, Lovekesh; and Medicherla, Raveendra Kumar.
The application for the patent was published on Sept. 12, under issue no. 37/2025.
According to the abstract released by the Intellectual Property India: "Automated translation of legacy code to modern programming languages is a need of the hour for modernizing enterprise systems. Traditional rule-based tools perform statement-wise translation, overlooking possible modularization and refactoring of source code to translate to readable target code. However, these encounter difficulties with regard to logical correctness and readability during translation. Present disclosure provides a large language model-based system and method that leverage temperature sampling and refinement-based strategies, to ensure logical correctness of transpilation and maximize readability of the target code. More specifically, sequential refinement of the transpiled high-resource source code with execution-guided logic feedback is performed followed by application of LLM-based readability feedback/criteria to yield better performance in terms of logical correctness and readability."
Disclaimer: Curated by HT Syndication.
