Music recommending method, music recommending device and music recommending system

Abstract

The invention discloses a music recommending method, a music recommending device and a music recommending system. The music recommending method comprises the following steps of: step S1: obtaining visit records of users on the music; Step S2: searching labels of the music in the access records; step S3: obtaining user features of the users through the labels; Step S4: clustering a plurality of users into user groups in the preset quantity according to the user features, wherein each user group has the respective user group feature; and step S5: recommending the music corresponding to the user group to the users in the user groups. The music recommending method, the music recommending device and the music recommending system have the advantage that the music recommending precision and the recommending speed can be improved.

Description

technical field [0001] The present invention relates to the field of information processing, in particular to a music recommendation method, device and system. Background technique [0002] At present, music recommendation technology can be summarized into two types: one is recommendation based on user behavior; the other is recommendation based on user group. [0003] Recommendation based on user behavior refers to making recommendations based on the user's past behavior. The basic principle is to obtain the target user's access log, analyze and organize the access log to obtain the user's behavior habits, and then perform fuzzy query in the music database according to these behaviors, and recommend the query result to the target user. This kind of recommendation technology based on user behavior is only a simple fuzzy query based on user behavior, without considering group information. The recommendation efficiency and recommendation range are relatively low, resulting in...

AI Technical Summary

Problems solved by technology

[0005] First of all, the above-mentioned user group-based recommendation technology only considers the information of similar user groups, but ignores the music content information, that is, if other users in the user group where the target user belongs to have heard a piece of music, they will directly recommend the music to the target user. users, without considering the content information of music, resulting in low recommendation accuracy and even wrong recommendation

[0006] Secondly, a certain music can only be recommended to other users if the user has taken actions (such as browsing, downloading, auditioning, evaluating, etc.). If the song is newly listed, it will not be recommended, resulting in a narrow recommendation range. question

[0007] Again, the similarity between users is obtained by comparing the historical behavior records of the target user with the historical behavior records of all other users. It takes a lot of time to make millions of comparisons, and with the expansion of the user scale, the time required increases exponentially, which leads to the inefficiency of the recommendation service

Images