摘要
在应用下载网站上,当用户浏览或下载某应用时,说明他对这个应用产生一定兴趣,如果网站显示和当前应用类似的其他应用,那么可以增加用户继续下载应用的概率。将基于行为推荐和基于内容推荐结合成推荐模型,并基于Hadoop进行实现,可以处理较大量的数据并得到较优的推荐结果。
When users are browsing or downloading one application on the application downloading website, it indicates that they are interested in the application. If there has other similar applications displayed for the user, there is a big possibility that the he will download them as well. A recommendation model which combined by the user-based and item-based recommendation, and implement it using the map and reduce method in Hadoop can process the big data and get the satisfying results.
出处
《现代计算机》
2013年第17期20-23,共4页
Modern Computer
关键词
关联推荐
余弦相似度
置信度
提升度
Associate Recommendation
Cosine Similarity
Confidence
Lift