基于项目特征聚类的协同过滤推荐算法  

A COLLABORATIVE FILTERING ALGORITHM BASED ON CLUSTERING OF ITEMS'CHARACTER

作　　者：翁小兰[1] 庄永龙[2] 

机构地区：[1]淮阴师范学院计算机科学系,江苏淮安223300 [2]淮阴师范学院信息中心,江苏淮安223300

出　　处：《计算机应用与软件》2009年第7期260-262,共3页Computer Applications and Software

摘　　要：提出基于项目特征聚类的Item-based协同过滤推荐算法。该算法首先根据项目的属性特征对项目进行聚类,形成其特征相似群,然后采用一种基于预评分的相似性度量方法计算目标项的最近邻居,最终产生推荐。经实验验证该算法可以有效解决用户评分数据稀疏性和冷启动的难题,而且可以显著提高系统推荐质量。An item-based collaborative filtering algorithm based on the clustering of the items＇character is proposed in this paper. Firstly it applies clustering method to items according to the main attributes of items and gets their character similarity group, then it calculates nearest neighbour of items with a similarity metric approach based on pre-rating, and finally it makes the recommendation. The experimental resuhs validated that this algorithm can improve the extreme sparsity of user rating data and the ＂cold-stat＂problem efficiently, and significantly raise recommendation quality.

关 键 词：协同过滤推荐 聚类 推荐系统 MAE 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TP751[自动化与计算机技术—计算机科学与技术]