Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{201870,
        author = {Vaishnavi Satish Deshmukh and Samruddhi Shankar Mote and Suraj Suresh Divate and Mr. Jagtap Kiran Prakash},
        title = {AI-Based Logical Error Detection and Explainable Code Analysis using Graph Neural Networks and Large Language Models},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {6831-6837},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201870},
        abstract = {S This paper introduces a Hybrid AI System that tackles logical error detection and explainable code analysis to make debugging smarter and help developers actually understand their code. The setup uses both Graph Neural Networks (GNNs) and Large Language Models (LLMs), which work together to dig into Python code. Here’s how it goes: first, the system breaks down the code into Abstract Syntax Trees (ASTs) and logical flow graphs. That way, it grabs not just the structure, but how things play out when the program runs.
A GNN, trained on piles of code—both correct and flawed—hunts for sketchy patterns and logical slip-ups. When it finds something odd, the LLM jumps in, offering explanations you can actually read, plus suggestions and corrected code. There’s more: the system builds interactive flowcharts and graph visuals, with those buggy parts marked bright red so you don’t waste time hunting them down.
All the debugging action happens inside a clean, web-based interface. It’s not just about catching errors—it helps you follow the code’s logic, pin down where things went wrong, and see it visually. With all these layers working together, debugging gets faster, code insights deeper, and the whole process more transparent. It’s a hands-on, AI-powered approach for program analysis and spotting logical errors right when they show up.},
        keywords = {Abstract Syntax Tree (AST), Automated Debugging, Control Flow Graph (CFG), Explainable Artificial Intelligence (XAI), Graph Neural Networks (GNN), Large Language Models (LLM), Logical Error Detection, Program Analysis, Source Code Analysis.},
        month = {May},
        }