Improved Hybrid Deep Collaborative Filtering Approach for True Recommendations

Recommendation services become an essential and hot research topic for researchers nowadays. Social data such as Reviews play important role in the recommendation of products. Improvement was achieved by deep learning approaches capturing user product information from a short text. However, previously used do not fairly efficiently incorporate users’ preferences characteristics. The proposed novel Hybrid Deep Collaborative Filtering (HDCF) model combines capabilities interaction modeling with high performance True Recommendations. To overcome cold start problem, new overall rating is generated aggregating Multivariate Rating DMR (Votes, Likes, Stars, Sentiment scores reviews) different external sources because sites have about same that make confusion to decision, either truly popular or not. HDCF consists four major modules User Product Attention, Filtering, Neural Classifier, (UPA-DCF + NSC DMR) solve addressed problems. Experimental results demonstrate our outperforming state-of-the-art IMDb, Yelp2013, Yelp2014 datasets true top-n products using increase accuracy, confidence, trust services.