A Multi-Objective Genetic Algorithm Framework for Efficient Association Rule Mining in Large-Scale Datasets

Abstract

Association Rule Mining (ARM) is a fundamental data mining technique for discovering interesting relationships in large datasets. However, traditional ARM algorithms face significant scalability and efficiency challenges when applied to big data contexts, often generating excessive redundant rules with diminished practical utility. This paper proposes a novel Genetic Algorithm (GA) framework specifically designed for large-scale ARM that employs multi-objective optimization. The framework simultaneously optimizes three key rule quality metrics—support, confidence, and lift—while explicitly minimizing rule redundancy and complexity. Experimental evaluation on multiple large-scale datasets demonstrates that our GA-based approach significantly outperforms conventional methods like Apriori and FP-Growth in both computational efficiency and rule quality. The proposed framework reduces execution time by up to 84% compared to Apriori and up to 42% compared to FP-Growth while increasing the proportion of high-quality, actionable rules by approximately 35% compared to traditional methods. These results confirm the effectiveness of our multi-objective GA framework as a robust and scalable solution for knowledge discovery in contemporary big data environments.