计算机应用与软件Issue(9)：293-295,3.DOI:10.3969/j.issn.1000-386x.2013.09.081
结合项目相似度的协同过滤算法
COLLABORATIVE FILTERING RECOMMENDATION ALGORITHM IN COMBINATION WITH ITEM SIMILARITY
徐文龙 1严泳键 2范冰冰1
作者信息
- 1. 华南师范大学计算机学院 广东 广州510631
- 2. 中山大学信息科学与技术学院 广东 广州 510006
- 折叠
摘要
Abstract
The increase of dataset sparsity leads to traditional collaborative filtering algorithm reducing its recommendation accuracy .To address this problem , a collaborative filtering recommendation algorithm in combination with item ’ s similarity is proposed .It calculates the similarity between the items first , and then predicts the rating valuations of the user unevaluated items according to this similarity in order to decrease the sparsity of dataset formed from the target users and their candidate nearest neighbours .Finally, according to user similarity , we get the item recommendation set .Experimental result shows that the algorithm can raise the accuracy of nearest neighbour search , therefore improves the recommendation quality of the collaborative filtering .关键词
推荐系统/协同过滤/项目相似度/预测估值/平均绝对误差Key words
Recommendation system/Collaborative filtering/Item similarity/Prediction valuation/Mean absolute error ( MAE)分类
信息技术与安全科学引用本文复制引用
徐文龙,严泳键,范冰冰..结合项目相似度的协同过滤算法[J].计算机应用与软件,2013,(9):293-295,3.
