| 基于矩阵划分和兴趣方差的协同过滤算法 |
| |
| 引用本文： | 潘红艳,林鸿飞,赵晶.基于矩阵划分和兴趣方差的协同过滤算法[J].情报学报,2006,25(1):49-54. |
| |
| 作者姓名： | 潘红艳 林鸿飞 赵晶 |
| |
| 作者单位： | 大连理工大学计算机科学与工程系,大连,116024 |
| |
| 摘 要： | 数据稀疏性是协同过滤系统面临的一个巨大挑战。本文提出了一种新的推荐算法———基于矩阵划分和兴趣方差的协同过滤算法。该算法采用矩阵分块的思想来缩小最近邻搜索的范围。矩阵分块时,采用聚类的方法,大大降低了矩阵的维度和稀疏等级。同时引入兴趣方差的概念,提高了计算最近邻的准确度。实验证明,本文提出的过滤算法在预测精度上较传统的推荐算法有很大的提高。
|
| 关 键 词： | 协同过滤 矩阵分块 兴趣方差 |
| 修稿时间： | 2005年3月24日 |
Collaborative Filtering Algorithm Based on Matrix Partition and Interest Variance |
| |
| Pan Hongyan,Lin Hongfei,Zhao Jing.Collaborative Filtering Algorithm Based on Matrix Partition and Interest Variance[J].Journal of the China Society for Scientific andTechnical Information,2006,25(1):49-54. |
| |
| Authors: | Pan Hongyan Lin Hongfei Zhao Jing |
| |
| Abstract: | Data sparseness is a serious problem in collaborative filtering system.In this paper,a new recommendation algorithm is presented,that is,a collaborative filtering algorithm based on Matrix Partition and Interest Variance.It partitions the huge matrix into some sub-matrixes in order to reduce the scale of searching nearest neighbors.In the course of partitioning the matrix,a clustering approach is applied to divide the sub-groups.Moreover,the concept of interest variance is adopted to improve the veracity of searching nearest neighbors.It proves that this method can obtain a better predictive precision,compared with traditional recommendation algorithm. |
| |
| Keywords: | collaborative filtering matrix partition interest variance |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
