Recommender system based on collaborative filtering
Abstract
Collaborative filtering is a popular technique for providing personalized recommendations in recommender systems. However, the sparsity problem and the accuracy-diversity tradeoff are major challenges that limit its performance. In this article, we propose a novel approach that combines matrix factorization with novelty metrics to improve the accuracy and diversity of recommendations. We evaluate our approach on the MovieLens dataset and compare it with several state-of-the-art techniques, including neighborhood-based methods, probabilistic models, and hybrid approaches. Our experimental results show that our method is better than other techniques in terms of both accuracy and diversity, as measured by precision, recall, and novelty metrics.
References
2. Hardesty, L. (Nov. 2019). The history of Amazon’s recommendation algorithm: Collaborative filtering and beyond, in ‘Amazon Science’, Amazon [Electronic resource]
3. Musa, J.M.; Zhihong, X. (2020). Item Based Collaborative Filtering Approach in Movie Recommendation System Using Different Similarity Measures. ICCTA '20: Proceedings of the 2020 6th International Conference on Computer and Technology Applications (pp. 31-34).
4. Schedl, M., & Knees, P. (2018). Music recommendation and discovery in the age of streaming services. Proceedings of the IEEE, 106(4), 626-641.
5. Choi, Jeongwhan; Hong, Seoyong; Park, Noseong; Cho, Sung-Bae (2022). "Blurring-Sharpening Process Models for Collaborative Filtering".
Abstract views: 0 PDF Downloads: 0