Research And Implementation On Music Recommender System Based On LFM

With the rapid development of the Internet,the number of digital music has the terrible rapid increase,users dig out their interest song from the massive digital music information is a very difficult thing,there has an "information overload" problem.The music recommendation system is an effective way to solve this problem.Traditional music recommendation methods are mostly based on the user's history ratings to establish a binary model to predict,while ignoring the time situational factors for the impact of music recommendations.Use the time context information to optimize the music recommendation to improve the user's satisfaction with the recommended results is a problem to be solved.The main work of this paper is to study the scheme that can be used to discover and use context information.The research focus on how to combine the time situation and analyze the user's historical behavior by time period.This paper first analyzes some user's organic relationship between song's favor degree with the time situation on the existing music recommendation platform,and calculates the user's song interest preference in the time period that obtaining by the recommended time.Based on this,the user's interest preference that in different time period is introduced,and improves the traditional LFM implicit semantic model,and the implicit semantic algorithm(TLFM)is established.For some users who's interest preference are more stable in certain time period,the immediate recommendation weight given to such users is based primarily on the preferences of this time period in history rather than the preferences based on all time,combined with user's instant behavior,given a reasonable recommendation.Secondly,through the experimental analysis,the TLFM algorithm is compared with the traditional LFM algorithm and the user-based cooperative filtering algorithm on the mean absolute error,the accuracy and the root mean square error on the the two data sets of Last.fm and DoubanFM,proves the algorithm designed in this paper.Finally,according to the functional requirements of the music recommendation platform,combined with the improved algorithm,give the detail design about the system architecture,function modules and database and use the MVC development model,combined with the improved TLFM algorithm to realize the music recommendation prototype system.The function test of the system is carried out,shows the recommended effect and verifies the successful application of the theory to practice.