A recommender system based on collaborative filtering using ontology and dimensionality reduction techniques. (February 2018)

• Record Type: Journal Article
• Authors: Nilashi, Mehrbakhsh; Ibrahim, Othman; Bagherifard, Karamollah
• Is Part Of: Expert systems with applications. Volume 92(2018)
• Page Start: 507
• Page End: 520
• Publication Date: 2018-02
• Subjects: Recommender systems -- Ontology -- Clustering -- Dimensionality reduction -- Scalability -- Sparsity; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2017.09.058
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A new method is developed for recommender systems. The recommender system is developed based on collaborative filtering. Scalability and sparsity issues in recommender systems are solved. MovieLens and Yahoo! Webscope R4 datasets are used for method evaluation. The method is effective in solving the sparsity and scalability problems in CF. Abstract: Improving the efficiency of methods has been a big challenge in recommender systems. It has been also important to consider the trade-off between the accuracy and the computation time in recommending the items by the recommender systems as they need to produce the recommendations accurately and meanwhile in real-time. In this regard, this research develops a new hybrid recommendation method based on Collaborative Filtering (CF) approaches. Accordingly, in this research we solve two main drawbacks of recommender systems, sparsity and scalability, using dimensionality reduction and ontology techniques. Then, we use ontology to improve the accuracy of recommendations in CF part. In the CF part, we also use a dimensionality reduction technique, Singular Value Decomposition (SVD), to find the most similar items and users in each cluster of items and users which can significantly improve the scalability of the recommendation method. We evaluate the method on two real-world datasets to show its effectiveness and compare the results with the results of methods in the literature. The results showed that our method is effective in improving the sparsity and scalability problems in CF.