Key quality characteristic selection incorporating customer attention under imbalanced data for popular products
研究了在电商非平衡数据下，如何将客户关注融入关键质量特性选择，提出自适应混合鲸鱼优化算法，以提升产品受欢迎程度。
Effective key quality characteristic (KQC) selection is essential for follow-up quality improvement. Customers’ demands for different product QCs affect product popularity. However, little research has integrated customer attention into KQC selection under imbalanced data in e-commerce, which can lead to a follow-up product with good quality but no popularity. This study, therefore, investigated KQC selection incorporating customer attention in the scenario of imbalanced data for popular products. First, KQC selection incorporating customer attention was defined as a multi-objective problem, aiming to minimise the percentage of selected QCs and maximise the importance of QCs, as well as cumulative attention to selected QCs. A collaborative filtering algorithm-based method was applied to extract customer attention from historical data when filtering key QCs. Second, an adaptive hybrid whale optimisation algorithm (AHWOA) was proposed to solve KQC selection. Here, simulated annealing was incorporated into the WOA agent search, and an adaptive convergence-acceleration mechanism and a fast non-dominated sorting algorithm with an improved crowding-distance measure were integrated into WOA. Third, the proposed AHWOA was evaluated on five datasets from the UCI repository, and the results show AHWOA’s advantages over five existing benchmark algorithms.