USER FEATURE GENERATING METHOD, DEVICE, AND APPARATUS, AND COMPUTER-READABLE STORAGE MEDIUM

A user feature generating method, device, and apparatus, and a computer-readable storage medium, relating to the field of video recommendation. The user feature generating method comprises: acquiring n groups of timing correspondences between target videos and user accounts, each group comprising each user account that has viewed the target video, and sorting the user accounts according to viewing timestamps, wherein n is a positive integer (201); acquiring a word embedding matrix by means of mapping according to the n groups of timing correspondences, wherein the word embedding matrix comprises a word vector corresponding to each user account (202); using a loss function to train the word embedding matrix, wherein the loss function is used to mine for a similarity relationship between the user accounts according to the degree of similarity of viewing histories (203); and determining a word vector corresponding to the user account in the completely-trained word embedding matrix to be a user feature of the user account (204).