An experimental comparative study of FP-growth based music recommendation algorithms

At present, the Internet products for the study of user interests and hobbies data mining and then recommended to increase user retention and user stickiness behaviour has been a common practice, music hobbies as an important point of interest for thousands of users, and its recommendation has become an important content of social life and has been the subject of more and more attention by the researchers, based on the collaborative filtering algorithm of the music recommendation is the use of the particularly wide range of recommendation algorithms. Music recommendation based on collaborative filtering algorithm is one of the most widely used recommendation algorithms, which mainly analyses the similarity between users or items, and completes the recommendation process based on this, but it has the problems of poor recommendation diversity and low coverage. Therefore, this paper designs comparative experiments to compare the recommendation algorithms based on FP-growth and collaborative filtering, relying on Last.fm's real user data warehouse, and demonstrates through experimental comparative research that compared with the collaborative filtering recommendation algorithm, the results of the recommendation content based on the FP-growth algorithm have a certain degree of enhancement in terms of diversity and coverage, and can produce a certain degree of accuracy in terms of the recommendation of music. accuracy, it can produce certain reference value.