论文部分内容阅读
随着微博和移动互联网的快速发展,社交网络变得越来越庞大,信息的个性化服务越来越受到重视.Slope One算法是一种协同过滤推荐算法,算法通过用户项目的偏差矩阵和频度矩阵,预测目标用户感兴趣的项目,实现了信息的个性化推荐.考虑到Slope One算法对大数据处理不足的问题,提出一种改进的基于并行Slope One算法的微博信息推荐算法,并借助MapReduce平台设计实现了该算法.实验表明,该算法不仅具有良好的加速比和可扩展性,还具有较好的预测准确率,它可以更高效的处理微博数据.
With the rapid development of microblogging and mobile Internet, the social network becomes more and more large, personalized information service is more and more attention.Slope One algorithm is a collaborative filtering recommendation algorithm, the algorithm through the user project deviation matrix and Frequency matrix to predict the target users interested in the project, to achieve the personalized recommendation of information.Considering the Slope One algorithm for large data processing problems, this paper proposes an improved algorithm based on parallel Slope One microblogging information recommendation algorithm, The algorithm is designed and implemented with MapReduce platform.The experiments show that the algorithm not only has good speedup and scalability, but also has a good prediction accuracy, which can deal with the Weibo data more efficiently.