Open AccessDesign and Implementation of Music Recommendation System Based on Hadoop
Author(s) -
Yufeng Zhao,
Xinwei Li
Publication year - 2018
Publication title -
international journal of advanced network, monitoring, and controls
Language(s) - English
Resource type - Journals
ISSN - 2470-8038
DOI - 10.21307/ijanmc-2018-045
Subject(s) - computer science , scalability , recommender system , collaborative filtering , cluster analysis , interface (matter) , scheme (mathematics) , set (abstract data type) , data mining , information overload , database , distributed computing , information retrieval , operating system , machine learning , world wide web , mathematical analysis , mathematics , bubble , maximum bubble pressure method , programming language
In order to solve the problem of information overload of music system under large data background, this paper studies the design scheme of distributed music recommendation system based on Hadoop. The proposed algorithm is based on the MapReduce distributed computing framework, which has high scalability and performance, and can be applied to the calculation and analysis of off-line data efficiently. The music recommendation system designed in this paper also includes client, server interface, database and ETL operation, which can calculate a set of complete recommendation system from user operation end to server and data calculation. In order to improve the accuracy of the recommendation algorithm, this paper introduces k-means clustering algorithm to improve the recommendation algorithm based on user-based collaborative filtering.The experimental results show that the accuracy of the proposed algorithm has been significantly improved after the introduction of k-means.