Hybrid Recommender System based on Singular value decomposition using content features
NeuroQuantology · 2024
In recent era of information technology and big data, a huge volume of data explosion on web makes it difficult to choose any information or item which users really interested about it.To handling such challenges, recommender systems (RS) emerged as helpful tool for recommending such information or item to potentially interested users.Content based recommendation (CBR) and collaborative filtering is two most popular classes of recommendation.Collaborative filtering (CF) approach analyzes user interest while CBR recommends items by object analysis.The main challenges of CF include scalability, synonymy and sparsity which creates obstacle in effective recommendation.In this work, we propose hybrid model of recommender system based on singular value decomposition (SVD), which combines with CF and content based recommendation approach.Experimental evaluation is carried out via prediction accuracy metrics root mean squared error (RMSE) and mean absolute Error (MAE).Empirical results using real dataset demonstrates the effectiveness of proposed approach in comparison with others baseline traditional rating RS.Hybrid recommendation model by combined features shows improved quality and better accuracy in rating prediction, introduced a diversity factors in recommendation.