| CPC G06Q 30/0631 (2013.01) [G06F 16/637 (2019.01); G06F 16/639 (2019.01); G06F 16/65 (2019.01); G06F 16/683 (2019.01)] | 12 Claims |
|
1. A method for recommending music content by a server including a modeling module, a collection module, and a service module, the method comprising:
training, by the modeling module, a service model using training data and result data stored in a customer database for determining a spending pattern of music content based on a neural network learning;
training, by the modeling module, the service model using the training data and the result data stored in the customer database for estimating a music age based on the neural network learning;
obtaining, by the collection module, account information for an account of a user of a service which provides music content, the account information comprising a real biological age of the user of the account, the real biological age registered as personal information;
obtaining, by the collection module, based on the account information, content information and usability information, the content information comprising one or more properties of music content consumed by the account, and the usability information comprising a manner in which the music content is accessed;
estimating, by the modeling module, a music age corresponding to the account information, based on the account information, the content information, and the usability information, the music age comprising a virtual age matched with a spending pattern of music content of the user, the music age differing from the real biological age of the user of the account;
recommending, by the modeling module, additional music content based on the music age rather than the real biological age of the user of the account; and
causing, by a point of contact (POC) application, a device to play the additional music content instead of music content matching the real biological age of the user of the account;
wherein estimating, by the modeling module, the music age comprises:
outputting the spending pattern of music content corresponding to the account information by inputting the account information, the content information, and the usability information into the service model;
generating a first vector, a second vector, and a third vector from the account information, the content information, and the usability information, respectively;
generating a customer vector encoding the first vector, the second vector, and the third vector, the first vector encoding account information comprising the real biological age of the user of the account, the second vector encoding content information comprising a property of music content consumed by the account, the third vector encoding usability information comprising a type of POC used to access music content by the account;
inputting the customer vector to a neural network of the service model;
outputting, from the neural network, a plurality of probabilities comprising, for each music age candidate of a plurality of music age candidates, a probability that the account information corresponds to the respective music age candidate, the plurality of music age candidates preset according to the spending pattern of music content; and
selecting any one music age candidate from among the plurality of music age candidates having a highest probability as the music age corresponding to the account information;
wherein training the service model for estimating the music age based on the neural network learning comprises applying a first weight corresponding to the account information, a second weight corresponding to the content information, and a third weight corresponding to the usability information;
wherein outputting the spending pattern comprises outputting the spending pattern corresponding to the input account information, the content information, and the usability information;
wherein outputting the probabilities comprises outputting probabilities corresponding to the music age candidates corresponding to the spending pattern;
wherein the result data corresponds to a result of recruiting a reference group, for each real age group of a plurality of real age groups, by a certain volume and analyzing spending patterns of users for music content; and
wherein the modeling module estimates the music age of all users based on the spending pattern of the reference group.
|