An Intelligent Hybrid Neural Collaborative Filtering Approach for True Recommendations

Recommendation services become a critical and hot research topic for researchers. A recommendation agent that automatically suggests products to users according their tastes or preferences instead of wandering in huge corpus product. Social data such as Reviews play an important role the products. Improvement was achieved by neural network methods capturing user product information from short text. However, approaches do not fairly efficiently incorporate users’ characteristics. We are proposing novel Hybrid Neural Collaborative Filtering (HNCF) model combines deep learning capabilities interaction modelling recommender systems with rating matrix. To overcome cold start problem, we use new overall aggregating multivariate MR (votes, likes, stars sentiment scores reviews) different external sources because sites have about same The propose consists four major modules HUAPA-DCF+NSC+MR (Hierarchical User Attention Hierarchical Product Attention, Deep Filtering, Sentiment Classifier, multivariant rating) solve addressed problems. Initially, HUAPA module is based on BiLSTM’s hierarchical attention (HUA) (HPA) embed characteristics respectively. Further, these combined nonlinear representations fed input module. Secondly, collaborative filtering implemented find explicit between Thirdly, NSC will extract user’s semantic incorporating Finally, uses (multivariant maximum extent final classification. Experimental results demonstration our outperforming than state-of-the-art at IMDb, Yelp2013 Yelp2014 datasets true top-n using HNCF (HUAPA+DCF+NSC+MR) increase accuracy, confidence, trust services.