MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202611088671 A) filed by Manipal University Jaipur on July 21, 2026, for A Cross-Language Software Vulnerability Detection System Using Unified Code Property Graphs And Patch-Aware Graph Neural Networks.
Inventor includes Dr. Manish Joshi.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention relates to a software vulnerability detection system for identifying security vulnerabilities in source code using artificial intelligence and graph- based program analysis. The system comprises a source code parsing module for generating abstract syntax trees from source code written in multiple programming languages, a graph construction module for generating a unified Code Property Graph (CPG) by integrating syntax, control flow, and data dependency relationships, and a multilingual code embedding module for generating semantic representations of program elements. A graph neural network processing engine analyzes the unified Code Property Graph using contrastive learning trained on vulnerable and patched source code pairs to detect software vulnerabilities. An explainability module identifies graph regions contributing to vulnerability predictions and generates remediation information, while a deployment module integrates the system with continuous integration and continuous deployment (CI/CD) pipelines for automated code analysis. The system further supports continuous model refinement using newly identified vulnerabilities and software patches, thereby improving detection accuracy and adaptability across multiple programming languages.
Disclaimer: Curated by HT Syndication.
