Analysis and evaluation of algorithms for personalization of interaction with the user for the development of a social network

This article analyzes personalization algorithms for social networks, with key objectives being the enhancement of user interaction and the improvement of recommendation relevance. The goal of this work is to evaluate various personalization approaches, such as recommendation systems and machine learning algorithms, as well as to assess the accuracy of these algorithms. Personalization approaches based on recommendation systems and machine learning methods are discussed, along with the application of artificial intelligence to improve recommendation accuracy. Three primary recommendation system algorithms are presented: collaborative filtering, content-based filtering, and hybrid models. Collaborative filtering was selected as the main personalization method, using the Python Library Surprise, which includes algorithms such as Singular Value Decomposition, Slope One, and K-Nearest Neighbors. A comparative analysis of Root Mean Squared Error and Mean Absolute Error metrics revealed that the K-Nearest Neighbors algorithm showed the best results, making it the preferred choice for further implementation. The final model, trained on the full dataset, demonstrated strong accuracy and potential for practical use in real products. The results presented could be valuable for social network developers in choosing optimal algorithms to enhance user experience, as well as for future research in personalization and recommendation systems.