Recommendation systems are essential for providing personalized content and improving user interaction on digital platforms. This research introduces a hybrid method that merges content-based filtering with collaborative filtering, utilizing matrix factorization to enhance recommendation accuracy. Contentbased filtering (CBF) analyses item characteristics are used to recommend related products, while collaborative filtering (CF) uncovers patterns in user-item interactions. However, handling large datasets poses a scalability challenge. To address this, the user-item rating matrix is expressed as a bipartite graph, and the Louvain algorithm is applied for community detection. This method improves predictions by focusing on localized patterns rather than the entire dataset. Additionally, pearson similarity is used for CBF, while matrix factorization enhances collaborative filtering leading to more accurate recommendations. The system’s effectiveness is evaluated using Root Mean Squared Error, Mean Squared Error, and Mean Absolute Error on MovieLens100 K, MovieLens-1M, and Anime Recommendation datasets. It demonstrates that integrating community detection with hybrid filtering enhances both scalability and accuracy, making the system more effective for real-world applications.