基于Spark的Hybrid推荐算法的研究与实现  

Research and implementation of Hybrid recommendation algorithm based on Spark

作　　者：祝永志 Zhu Yong-zhi(School of Information Science and Engineering,Qufu Normal University,Rizbao 276826,China)

机构地区：[1]曲阜师范大学信息科学与工程学院,日照276826

出　　处：《电子技术（上海）》2018年第12期59-62,共4页Electronic Technology

基　　金：山东省自然科学基金(ZR2013FL015);山东省研究生教育创新资助计划(SDYY12060)

摘　　要：协同过滤推荐技术作为推荐系统的一个重要分支,成为目前应用最广泛的一种推荐算法。但协同过滤算法仍然显露出数据稀疏性问题、可扩展性问题等。为解决上述问题,文章提出了基于Spark平台的基于ALS算法和物品相似度相结合的混合协同过滤算法。本文算法在一定程度上解决了因数据量不足带来的数据稀疏性问题,基于Spark分布式并行计算框架技术也解决了可扩展性问题,同时又提高了算法推荐的准确性。基于Movielens数据集的实验表明,文章算法具有可扩展性高、响应时间短以及推荐精度高等特点。Collaborative filtering recommendation technology, as an important branch of recommendation system, has become the most widely used recommendation algorithm at present. But the collaborative filtering algorithm still shows the problem of data sparsity, scalability and so on. To solve the above problems, based on the Spark platform, a hybrid collaborative filtering algorithm based on ALS algorithm and item similarity is proposed in this paper. This algorithm solves the problem of sparse number caused by insufficient data, and the distributed parallel computing framework based on Spark also solves the scalability problem, and also improves the accuracy of the algorithm recommendation. Experiments based on Movielens dataset show that the algorithm has the characteristics of high scalability, short response time and high recommendation accuracy.

关 键 词：Hybrid推荐算法 ALS SPARK 可扩放性 协同过滤. 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]