摘要
协同过滤算法是根据基本用户的观点产生对目标用户的推荐列表,现模拟蚂蚁觅食的原理,将用户视为具有不同属性的蚂蚁,聚类中心视为蚂蚁所要寻找的"食物源",提出基于蚁群算法实现用户聚类,以提高协同过滤推荐系统的最近邻查询速度,降低搜索开销,同时避免了使用K-Means聚类方法受初始聚类中心和聚类个数的影响。最终实验验证蚁群算法实现用户聚类的有效性,且解决了新用户得不到推荐的问题,并提高了协同过滤推荐算法的精确度。
Collaboration filtering recommendation algorithm is that generate the recommendation list according to basic user＇ view. Now imitated ant foraging theory, the users are regarded as different attributes ants, clustering center is regarded as the ＂food source＂ that the ants are looking for, proposed to cluster user based ant algorithm, for improving the query speed of the nearest neighbor in the collabora- tive filtering recommendation system, reducing the search spending, and avoiding the effects of initial clustering centers and clustering numbers in the use of K-Means clustering method. Finally, the experiment verify that user clustering through ant algorithm is effective, and solve the problem of new user not recommended, enhance the precision of collaboration filtering recommendation algorithm.
出处
《计算机技术与发展》
2011年第10期73-76,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(70871081)
上海市重点学科建设资助项目(S30504)
关键词
蚁群算法
聚类
协同过滤
推荐
用户
ant algorithm
clustering
collaboration filtering
recommendation
user
