Mining Skyline Frequent-Utility Pattern with Threshold Filtering
Jimmy Ming‐Tai Wu, Yadong Liu, Huiying Zhou, Fengyang Li, Shaowei Ma · 2023
In the past, skyline pattern mining algorithms have achieved extensive research. The current algorithms for mining skyline patterns are inefficient and time-consuming. In practical applications, they may be inadequate because traditional skyline patterns can result in points that are extremely biased towards one dimension. Therefore, this paper introduces a novel approach for mining skyline patterns that incorporates threshold settings. Unlike traditional methods, this new approach uses multiple thresholds to eliminate extreme points that don't meet the user's requirements in certain dimensions. By doing so, the number of candidate search items is greatly reduced, resulting in a more efficient and effective search process. The proposed algorithm was evaluated through extensive experiments on various databases, and its performance was compared with that of popular existing skyline pattern mining algorithms. The results demonstrate that the new approach not only yields a more streamlined and optimized set of results, but also outperforms previous algorithms in terms of both runtime and search space.