Research and Application of a Dual Filtering Music Hybrid Recommendation Model Based on Cat Boost Algorithm and DCN

Abstract

With the increase of Internet users, the traditional music recommendation model can not meet the increasing personalized needs of users. The single deep cross network model has some defects in music recommendation, such as poor stability and inability to process complex data. To overcome the shortcomings of existing models, a new hybrid music recommendation model combining CatBoost algorithm and deep cross network is constructed to improve the recommendation performance and better meet the individual needs of users. Then the performance of the hybrid model is compared with other algorithms. The results showed that the accuracy of the proposed hybrid algorithm was up to 92.7%, which was superior to the comparison algorithm. In comparison with other single model and hybrid model, it is found that the proposed model was more than 0.05% higher than other models in the four indices of AUC area, accuracy, precision and recall. The above results showed that the proposed hybrid music recommendation model could efficiently process data information and provide users with accurate personalized music recommendation. This study not only promoted the development of music consumption and creation, but also found that the CatBoost-DCN hybrid model was significantly effective in improving recommendation performance. This finding provides a more efficient recommendation strategy for music platforms and has far-reaching significance for improving user experience and satisfaction.