Approaches to Integration of Fuzzy Logic in User-Based Collaborative Filtering Methods

The article analyzes approaches to integrating fuzzy logic into user-based collaborative filtering methods to address the problems of data sparsity and uncertainty in user preferences. It emphasizes that collaborative filtering is a popular tool for implementing recommendation systems, which heavily relies on the accuracy of numerical data representation and the completeness of information during the recommendation process. The use of fuzzy logic enables modeling under ambiguity and uncertainty through the use of linguistic variables and membership functions of fuzzy values. Examples of systems successfully integrating this approach are examined, combining fuzzy logic with collaborative filtering methods, the use of fuzzy association rules, and multi-level similarity, which allows for effective handling of sparse and incomplete data. A mathematical framework is proposed for constructing genre profiles of users and films, which supports recommendation personalization. Gaussian-like membership functions and the application of cluster analysis allow for accounting for multi-genre content and the fuzziness of user preferences. The proposed algorithms reduce the impact of individual biases and ensure recommendation accuracy even under conditions of fragmented data. The research results demonstrate the potential of hybrid approaches in developing recommendation systems for various domains, including online stores, streaming platforms, and project management. Further testing of the proposed methods opens up opportunities for their large-scale application