Title: Artificial Intelligence Across Disciplines: Research, Innovation, and Intelligent Solutions
Editors: Dr. Subita Bhagat, Dr. A. Balamurugan, Dr. P. Krishna Kumar, and Mrs. S. Nandhini Devi
ISBN: 978-81-69857-83-3
Chapter: 9
DOI: https://doi.org/10.59646/815/09
Author: Dharani M
Abstract
Artificial Intelligence for Software Engineering (AI4SE) has evolved from rule-based static analyzers to deep representation learning frameworks that model semantic, structural, and control-flow dependencies in source code. This chapter examines the design, mathematical modeling, and operational evaluation of an end-to-end intelligent software engineering pipeline applied to enterprise microservice ecosystems. By integrating abstract syntax trees (ASTs), inter-procedural control flow graphs (CFGs), and data dependency graphs (DDGs) into a unified Code Property Graph (CPG), we formulate a relational graph neural network with multi-head self-attention to support automated vulnerability discovery, test suite optimization, and self-healing patch synthesis. Evaluated across a longitudinal case study of 1.4 million lines of Java and C++ code, the framework achieves a defect detection Matthews Correlation Coefficient (MCC) of 0.814, accelerates test execution by 64.2% via predictive test prioritization, and reduces mean time to remediation (MTTR) by 58.7% through automated patch generation.
Keywords: Artificial Intelligence, Software Engineering, Code Property Graph, Graph Neural Networks, Automated Program Repair, Predictive Testing