Collaborative Optimization of Frequent Itemsets Mining with Game Theory and Heuristic Algorithms

Traditional frequent itemset mining algorithms often suffer from high time complexity and memory bottlenecks when processing large-scale datasets, which limits their practical performance in frequent itemset mining applications. To address this issue, this paper proposes a novel frequent itemset mining method that integrates game theory with a genetic algorithm, called GameGA-Apriori. First, by optimizing the initialization, crossover, and mutation mechanisms of the GA-Apriori algorithm, population diversity is enhanced, redundant computations are minimized, and mining efficiency is improved. Second, game theory is innovatively incorporated into frequent itemset mining, where binary codes are distributed among game players, and rules are generated through strategic competition and cooperative optimization. To mitigate the issue of premature convergence, which tends to arise when the number of game players or the initial population size exceeds a critical threshold, a correction strategy is proposed. This strategy dynamically adjusts algorithm parameters and the evolutionary mechanism, significantly improving both the quantity and quality of the mining results, particularly in high-dimensional data scenarios. Experimental results demonstrate that the optimized algorithm achieves a broader mining range and higher accuracy across various initial population sizes. Overall, this study introduces a novel approach for optimizing frequent itemset mining, verifies the feasibility and effectiveness of integrating game theory into heuristic mining frameworks, and lays the groundwork for further research in high-dimensional data mining.