| 基于融合GMM聚类与FOA-GRNN模型的推荐算法 |
| |
| 作者姓名： | 李毅鹏 阮叶丽 张杰 |
| |
| 作者单位： | 中南财经政法大学信息与安全工程学院，湖北 武汉 430073 |
| |
| 基金项目： | 校级中央高校基本科研基金资助项目;教育教学改革基金资助项目(2018-9) |
| |
| 摘 要： | 针对传统基于物品的推荐算法由于数据稀疏性导致的低推荐精度问题，提出了一种融合GMM聚类和FOA-GRNN模型的推荐算法。该算法首先使用高斯混合模型（GMM）方法对物品特征进行聚类；然后根据聚类结果分别构造评分矩阵，并使用Slope One算法填充评分矩阵；最后计算用户对物品的相似度预测评分作为输入，通过FOA-GRNN模型输出最终的评分。基于movielens-2k数据集的实验结果表明，与其他3种算法相比，该算法能够更好地处理高稀疏性数据，推荐精度更优，并能够在一定程度上解决冷启动问题。
|
| 关 键 词： | 推荐算法 GMM聚类 果绳优化 广义回归神经网络 SlopeOne算法 |
Recommendation algorithm based on GMM clustering and FOA-GRNN model |
| |
| Authors: | Yipeng LI Yeli RUAN Jie ZHANG |
| |
| Affiliation: | School of Information and Safety Engineering,Zhongnan University of Economics and Law,Wuhan 430073,China |
| |
| Abstract: | Aiming at the problem of low recommendation accuracy caused by sparse data in traditional item-based recommendation algorithm,a recommendation algorithm based on GMM clustering and FOA-GRNN model is proposed.The algorithm firstly uses Gaussian mixture model (GMM) to cluster the item features,then constructs the score matrix according to the clustering results,and fills the score matrix with slope one algorithm.Finally,the user's score based on similarity prediction is taken as input,and the final score is output through FOA-GRNN model.Experimental results based on movielens-2k dataset show that the proposed algorithm can deal with highly sparse data better and has better recommendation accuracy than the other three algorithms. |
| |
| Keywords: | recommendation algorithm GMM clustering FOA GRNN Slope One algorithm |
|
| 点击此处可从《》浏览原始摘要信息 |
|
点击此处可从《》下载全文 |
