An Interactive Attention Network with Stacked Ensemble Machine Learning Models for Recommendations
Ahlem Drif, Saadeddine Selmani, Hocine Cherifi · 2022
This chapter proposes and investigates a hybrid recommender system where the recommended content is accurate and personalized for each user. Research around recommender systems falls into three main categories: collaborative filtering approaches, content-based approaches and hybrid approaches that combine the two techniques. The chapter introduces an interactive attention mechanism to model the mutual influence relationship between aspect users and items. The interactive neural attention network-based collaborative filtering recommender exploits the encoding ability of the interactive attention between users and items. The stacked content-based recommender is composed of a stack of machine learning models. The empirical evaluation demonstrates that the proposed framework of “Interactive Personalized Recommender” significantly outperforms state-of-the-art baselines on several real-world datasets. The chapter reviews the datasets, evaluation measures and alternative techniques used in the experiments. It presents the hyperparameters analysis step performed separately on each recommender.