基于Mahout的新用户推荐算法的设计与实现

Design and implementation of a new user recommendation algorithm based on Mahout

为了解决大数据背景下新用户因没有历史数据而导致推荐难和推荐效率低等问题,提出将基于Mahout的协同过滤算法与基于MapReduce的Top N算法相结合的技术方法,来实现新用户推荐算法,从而构建新用户推荐系统的架构,并对Hadoop Top N算法以及Mahout中协同过滤算法进行设计与实现。理论分析和实验验证表明,该新用户推荐算法在推荐效率、对大规模数据处理的伸缩性以及推荐质量上都明显优于单独使用协同过滤算法的新用户推荐。

Recommendation for new users in big data era is difficult and the efficiency is very low due to the lack of historical data. In order to solve these problems, we propose a new user recommendation algorithm, which combines the collaborative filtering algorithm based on the Mahout and the Top N algorithm based on the MapReduce. We build a new user recommendation system architecture, design and implement the Hadoop Top N algorithm and the collaborative filtering algorithm in the Mahout. Theoretical analysis and experimental results show that the proposed recommendation algorithm for big data processing has better recommended efficiency, scalability and quality than the collaborative filtering algorithm.