For Mobile Music Research And Improvement Of The Collaborative Filtering Algorithm

With the popularity of the rapid development of mobile communications networktechnology, and a lot of renovation of smart mobile devices, the Mobile E-commerce has asubstantial influence on people’s lives. Internet-based Mobile e-commerce has become animportant business model. The use of mobile devices for entertainment has been as part ofthe lives of many people. The mobile music has much variety, and there are mass data.That needs the recommended system to create high-efficiency, and make higher demandsto provide more accurate and personalized service. The collaborative filtering technologieshave a very wide range of applications in the field of electronic commerce, and haveachieved good results demonstration effect, but with the development use of mobilee-commerce, collaborative filtering techniques have some deficiencies in someapplications. For example, in some cases, they don’t consider to the characteristics ofmobile e-commerce, and less considering the characteristic attributes of the data in thecomplex environment and their relationships. Meanwhile, with the theoretical studies ofrecommendation systems, many researchers have started to pay attention to the practicalapplication of the recommendation algorithm in different environments. Especially, thereneed to consider the mobile music recommender system of the mobile environment. In thispaper, to solve this problem, I have studied the collaborative filtering algorithms and makeimprovements for the recommender system of mobile music. The designed algorithm isthat to Consider the characteristics of mobile e-commerce environment, and make acollection of the model of the relationship between users, and then to build the ratingsmatrix of the mobile music project. Then we use the improved model of personalizedinformation with the recommendation algorithm to combine with the user-based algorithmand the item-based algorithm.At the last part of the article, we make design an experiment and conduct theimplementation of the experiment. Then we make comparison and testing of the algorithms,using the error rate and accuracy that the recommended results to compare the improvedalgorithm and the original algorithms. It is proved that on the base of the combination bythe customer relationship model, the recommendation algorithm has achieved much betteraccuracy results than the traditional collaborative filtering algorithm.