Research On Music Recommendation System Incorporating Temporal Information

In the context of big data,the types and scales of data are increasing at an astonishing speed.Music resources are one of the enormous sources of data,and when faced with massive music resources,many users may feel overwhelmed.The emergence of music recommendation systems can help users cope with this problem by analyzing users’ historical behavior,interests,and other information to predict the music they may like and recommend corresponding music to them,helping users make choices and improving their usage experience.The commonly used algorithm in music recommendation currently is collaborative filtering recommendation algorithm,which finds the similarity and correlation between users through mining user behavior data and then recommends corresponding music to users.However,with the increase in data volume,collaborative filtering algorithms face more and more challenges in a massive data environment.This paper proposes a music recommendation algorithm that integrates time information to reduce the influence of time on music recommendation results and obtain more accurate recommendation results.The experimental results on the music dataset prove that the proposed music recommendation algorithm can significantly improve the quality of music recommendation compared with traditional recommendation algorithms,which is mainly reflected in precision,recall,F1 score,and other indicators.This also indicates that using reasonable and innovative algorithms can effectively improve the recommendation effect in the music recommendation system.To better apply the research results to practice,this paper designs and develops a music recommendation system based on the improved algorithm mentioned above.The main research work includes:1.Data preprocessing.This paper mainly processes the play counts in the music dataset by counting user play times to obtain a matrix representing user-music play frequency.By introducing nonlinear relationships,this paper further improves the accuracy of music rating results and combines the processed rating matrix with the researched time information function.2.Proposing a music recommendation algorithm that integrates time information.This paper first considers that users’ preferences for music are different at different times,so it introduces time effect functions and music season sensitivity functions and then combines them through weighted fusion for music recommendation.Experiments show that recommending in this way can reduce the bias caused by time information on music recommendation results to a certain extent and improve the accuracy of music recommendation results.3.Designing and developing a music recommendation system.This paper elaborates on the implementation method of the music recommendation system and applies the music recommendation algorithm proposed in this paper to the designed system.After module testing,the preliminary development of the system is completed.