A hybrid AI framework for hyper-personalized recommendations and customer engagement prediction in e-commerce

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The explosion of e-commerce has led to an increase in the demand for smart systems that can provide hyper-personalized experiences and enhance customer engagement. This study proposes a hyper personalized e-commerce framework by incorporating the deep learning (DL) based hybrid recommendation models along with the ensemble classifiers for engagement prediction. Real world public e-commerce datasets were used in a two-stage evaluation that measured recommendation effectiveness on Precision@10, Recall@10 and NDCG@10, and engagement prediction on Accuracy, F1-Score, and ROC-AUC. The results of the proposed framework were compared with the baseline approaches in terms of Precision@10, Recall@10, and NDCG@10, which gave Precision@10 = 0.382, Recall@10 = 0.441, and NDCG@10 = 0.469, Accuracy = 91.2%, F1-Score = 0.886, and ROC-AUC = 0.937. These results show that an ensemble of DL architectures and gradient boosting techniques strongly improves the quality of personalization and behaviour prediction. This study highlights the game changing power of recommender and predictive AI systems in driving customer experience and data-driven digital commerce approaches.