Hybrid Weight Factorization Recommendation System
Dineth Keshawa Jayathilake, Gayumi Nimesha Kottage, Kapuliyanage Chashika Chankuma, Gamage Upeksha Ganegoda, Thanuja Chandani Sandanayake · 2018
Various recommendation systems have been introduced to filter, prioritize and efficiently deliver relevant information to users of e-commerce sites. But each of these approaches contain disadvantages which are unique to them in performing recommendations. To cope up with this issue hybrid recommendation systems are introduced. This study introduces a novel weight factorization model to integrate multiple recommender models to overcome disadvantages such as data sparsity and cold start. The proposed hybrid model is evaluated against multiple recommendation models using Root-Mean-Squared-Error. Based on the experimental results that the proposed model performs better compared to content-based filtering, collaborative filtering and latent factor recommendation models.