Abstract

This paper proposes a novel hybrid framework that integrates soft computing techniques and machine learning algorithms to effectively address persistent optimization challenges in software engineering. Traditional ObjectOriented Programming (OOP) methodologies often struggle to manage modularity, coupling, and concern separation, particularly in large-scale systems. Drawing inspiration from aspect-oriented design principles, the proposed framework introduces a multi-layered approach that incorporates fuzzy logic, genetic algorithms (GAs), and neural networks to model imprecision, automate decision-making, and enhance prediction accuracy across software engineering tasks. The framework operates through five interconnected modules: input data acquisition from diverse software artifacts (e.g., source code, bug reports, version control history), data preprocessing, application of soft computing for optimization, machine learning for predictive analytics, and a decision support system that delivers actionable insights. Emphasis is placed on the selection and analysis of modularity-driven software quality metrics such as Coupling Between Objects (CBO), Lack of Cohesion in Methods (LCOM), and Class Dependency Analysis (CDA). Evaluation is conducted using real-world datasets from repositories like GitHub and PROMISE, supplemented by simulation-based testing to validate scalability and generalizability. Results demonstrate that the hybrid framework significantly improves maintainability, reusability, and software quality while reducing design complexity and development effort. The integration of advanced computational intelligence enables a more adaptive and accurate understanding of non-linear patterns in software processes, making it a robust tool for modern development environments. This study contributes to the advancement of intelligent software engineering tools through measurable performance enhancements.