Abstract: Recommender systems play a vital role in engaging a user on a platform by recommending similar new items based on users interest. Some of the tech giants like amazon use recommendation systems to recommend new items based on users previous search or purchases, Netflix uses recommendation systems to recommend new movies to users based on their interest. In past few years due to the availability of huge amounts of data, it is difficult to find information that is useful and relevant. Collaborative filtering, Content-based filtering, Hybrid filtering are some of the techniques used by recommendation systems. Few issues that exist in recommendation systems are data sparsity problem and cold start problem. This paper proposes a method to address the issue of sparsity in SVD based approaches. Data sparsity refers to problems in finding similar users because users only rate a few items. Cold start refers to difficulties in generating accurate recommendations for users who have rated very small items.
Keywords: Matrix Factorization (MF), Singular Value Decomposition (SVD), Sparsity, Collaborative Filtering (CF)
|
DOI:
10.17148/IJARCCE.2021.10455
