COLLABORATIVE FILTERING HYBRID RECOMMENDATION ALGORITHM INCORPORATING GAIFWFCM AND TFNS
金玉 1徐新卫 1陶飞 1韩业 2陈荣凯2
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 安徽工业大学管理科学与工程学院 安徽 马鞍山 243032
2. 安徽工业大学工程研究院 安徽 马鞍山 243032
折叠
摘要
针对传统推荐系统中使用离散评分未能合理表达用户偏好的问题,将遗传算法引入优化特征加权模糊 C均值,通过梯形模糊聚合相似目标用户,提出基于遗传算法的优化加权模糊 C 均值聚类融合梯形模糊数的协同过滤模型.通过遗传算法进行增强初始聚类中心,利用优化加权模糊 C 均值聚类融合梯形模糊数,分析类内与类间属性特征关系,实现用户精细划分,合理过滤推荐.在两组数据集中以 MAE 和 RMSE 作为评估指标进行实验,实验结果表明,所提算法在与其余 6 种算法对比中分类误差更低,对用户意愿识别更加清晰.
Abstract
In response to the issue of traditional recommendation systems' inability to effectively capture user preferences using discrete ratings,this study introduces a genetic algorithm to optimize feature-weighted fuzzy C-means.By employing trapezoidal fuzzy aggrega-tion to cluster similar target users,a collaborative filtering model based on the genetic algorithm optimized weighted fuzzy C-means clus-tering and trapezoidal fuzzy numbers is proposed.The genetic algorithm enhances initial clustering centers and utilizes weighted fuzzy C-means clustering and trapezoidal fuzzy numbers for fusing intra-class and inter-class attribute features,enabling fine-grained user seg-mentation and rational recommendation filtering.Experimental evaluations on two datasets utilizing MAE and RMSE as evaluation met-rics demonstrate that the proposed algorithm exhibits lower classification errors compared to six other algorithms while providing clearer identification of user preferences.
维普
万方数据