Role of Collaborative Filtering in Recommendation Systems

Abstract

Different categories of techniques can be utilized in recommendation systems (RS), like filtering by the content or hybrid filtering. However, the most extensive and popular RS is collaborative filtering (CF). Its main idea is to calculate and predict the user's interest in any item. If enough data is provided, CF -based RS is sufficient to provide the most accurate prediction as it is a user's preference-based technique. The most crucial part of RS is to predict its user's behavior, and in the past, user-based CF has done it successfully. But their specific usage has uncovered a few genuine problems, like information sparsity and information versatility, with, bit by bit increasing the number of clients and things. This work presents the central ideas of CF, its essential usage for clients with versatile networks, the hypothesis and practice for the calculation of CF and plan settlements concerning rating frameworks & appraisals securing.