MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641073844 A) filed by Vellore Institute Of Technology on June 15, 2026, for “an Early Defect Prediction System For Assessing Defect Risk From Early-Stage Software Development Artifacts”.
Inventors include Shikhar Tripathi; Shresth Chordia; and Dr. G Jagadeesh.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: The present invention relates to an early defect prediction system and method for assessing software defect risk from early-stage artifacts, including natural language requirements and UML class diagrams. The system analyzes requirements using a hybrid approach combining semantic representation from a pre-trained NLP model, a supervised classifier for defect probability estimation, and rule-based quality checks, while UML diagrams are parsed into graph form to extract structural metrics for design risk evaluation. A risk integration module combines requirement and design risk scores using weighted aggregation to generate a unified early defect risk score, which is then classified into a qualitative risk level, enabling early identification of potential software defects. Fig 1 to 3.
Disclaimer: Curated by HT Syndication.