.png){kind=icon}
.png){kind=icon}
Return
Hybrid fuzzy weighted NMF and NSGA-II optimization for a recommender system
.png){kind=icon}
PRE
AI
S
S
J
S
DOI：10.1007/s13042-026-03255-6.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The rapid growth of interactive data in recommender systems poses significant challenges in developing models that simultaneously ensure high accuracy and robustness for user preference prediction. To address this issue, we propose a collaborative filtering recommender system that introduces a novel hybrid algorithm. This algorithm integrates non-negative matrix factorization (NMF) with a newly designed weighting scheme tailored to establish a robust balance between prediction accuracy and model’s ability for generalization. To achieve high-quality multi-objective solutions, the Nondominated Sorting Genetic Algorithm II (NSGA-II) was employed. In the objective function formulation, the number of clusters (k) was optimized by considering a combination of objectives. This included prediction accuracy metrics such as mean absolute error, alongside clustering quality metrics like the Fuzzy Silhouette Index. This approach enables the simultaneous optimization of multiple objectives, leading to the discovery of a set of Pareto optimal solutions. The experimental evaluations on the MovieLens 100K, MovieLens 1M and Film Trust datasets demonstrated that the proposed approach yields robust and promising results across key performance metrics. These results demonstrate that the integrated framework of matrix factorization, clustering, and evolutionary optimization effectively enhances recommendation accuracy and stability within the investigated experimental scales.
Keywords:
Collaborative filtering
Fuzzy C-means
Clustering
Nondominated Sorting Genetic Algorithm II
Non-negative matrix factorization
Recommender system
Journal
IF:
2.7
Papers:
3.1K
Citations:
5.6K
Organization
Cited Papers