Gender Identification and Age Estimation of Users Based on Music Metadata.

Music recommendation is a crucial task in the field of music information retrieval. However, users frequently withhold their real-world identity, which creates a negative impact on music recommendation. Thus, the proposed method recognizes users’ real-world identities based on music metadata. The approach is based on using the tracks most frequently listened to by a user to predict their gender and age. Experimental results showed that the approach achieved an accuracy of 78.87% for gender identification and a mean absolute error of 3.69 years for the age estimation of 48403 users, demonstrating its effectiveness and feasibility, and paving the way for improving music recommendation based on such personal information.