Hybrid Attention-SVM Based Product Recommendation with Grey Wolf Optimization for E-Commerce Platforms

Abstract

With the rapid development of the e-commerce industry, personalized product recommendation models have received increasing attention. Traditional recommendation systems have shortcomings in capturing user interest features. This study proposes a product recommendation model based on a hybrid attention mechanism and Support Vector Machine (SVM), which makes recommendations more accurate and personalized. This model combines three attention mechanisms: Spatial attention automatically identifies the product image areas that users are concerned about; Channel attention dynamically adjusts the importance of feature channels to highlight the features that influence user decisions; Frequency attention optimization focuses on the detailed features of the product. Based on feature extraction, this study uses an SVM classifier for product recognition and classification and introduces a grey wolf optimization algorithm to adaptively adjust the core parameters of SVM, improving classification accuracy and robustness. The experimental results showed that the mean square error of the model was 0.19 in the training set and 0.07 in the validation set. Compared with the K-means clustering algorithm and backpropagation neural network, this algorithm has improved by 0.06 and 0.04. Meanwhile, the accuracy rate of personalized recommendation reached 0.702, which was 0.059 and 0.026 higher than that of K-means clustering and backpropagation neural networks. The operation time of the model was 1.04 seconds, demonstrating high practicability and efficiency. The research model has improved the depth and accuracy of feature extraction, consistent recommendation ability, and computational efficiency. This study provides a new practical personalized recommendation strategy for e-commerce platforms, which has broad application potential and economic value.