Design And Implementation Of Music Recommend System Based On Spark

After several years of development,Chang Ba mobile application as one of mainstream sing applications has accumulated much music,but there wasn't a personal recommendation system.It is difficulty for users find interesting music in such mass data.Traditional search engine is applied to the situation which user can use keywords describe his requirement.And the hot music list made by sort algorithm can not represent the interest of all users.Music is a typical item which has Long Tail appearance.Few of Music can be watched by users and much of it is' t cared by anyone.So personalization music recommendation system can help users find their insteresting music and achieve the goal of increasing usage of application.Recommendation system is one of information filter system.It predict the user's interest for some item by mining and analyzing the data of user and item deeply.At present,the popular recommendation algorithm includes Collaborative Filteringbased Recommendation and Content-based Recommendation.Collaborative Filtering-based Recommendation use the action data of user,otherwise Content-based Recommendation use the original data of item.Different recommendation algorithm has personal advantage and disadvantage.So recommendation system combine many algorithm for recommendating generally.This paper referring to current research result chooses four recomenndation algorithem to achieve in spark platform.The four algorithm are Collaborative Filtering-based Recommendation,Item Content-based combined with Item based Collaborative Filter Recommendation,Graph-based model Recommendation and Latent Factor-based model Recommendation.Because of requirement that recomenndationg is for many users and much music,we use spark platform to achieve the recommendation engine with corredponding algorithm.And more,we would concat the recommendation engine and business system to design and achieve the recommendation system which has complete flow path.