The Research On Technologies Of Music Recommending Based On Social Tags

With the rapid development of Internet technology,the pursuit of music has been greatly promoted by digital multimedia technology.Many music social networking sites are favored by public.Users are being buried in mass information due to rapid increase in music data.Meanwhile,users and suppliers are being challenged by information overloading.Traditional music search only adapted to users who have a clear goal and can use the key words to express what they want.Without considering the differences of the users,only a small part of the music searched are followed or downloaded while the majority part are ignored,which is a typical phenomenon called music long tail.For questions above,personalized music recommendation technology is promoted to provide effective solutions to the foregoing issues.In this paper,based on social tags,we had a further study the way to improve music recommendation results from the following aspects.Firstly,in order to improve the accuracy and diversity association rules based social tagging music recommendation method was proposed.Weight-labelled user interest model was established by combining the user of explicit and implicit feedback information.Then,by combing the User Interest Model and Association Rules F-Apriori Algorithm,higher music tags' results were recommended to the users.As users have to afford extra payment burden with the utilization of explicit feedback and are unable to feed back their true interest real-timely,which resulted in data noise,combing the implicit feedback to analyze users' preferences were considered and proposed Multiple Correspondence Analysis to remove the noise of tags.Secondly,for information sparsity and the cold start problem,a hybrid music recommending method based on social tagging and item attributes is proposed.As users' attribute information tends to be sparse due to privacy protection,item attributes was explored in this paper.With the help of implicit dirichlet model and item attribute,we classify new items and put new items with social tagging,and eventually made hybrid music recommendation with user interest.The hybrid recommendation method,which based on the socialization of tags and attributes of the objects in this paper,provides users with better personalized music recommendation service.