| CPC G06F 21/6254 (2013.01) | 18 Claims |
|
1. A method, in a data processing system, for customizing an avatar in a virtual environment, the method comprising:
identifying a first user that is interacting with the virtual environment and to whom the avatar is to be rendered in the virtual environment;
determining a classification of the first user with regard to a plurality of affinity groups specifying a level of affinity between the first user and an operator of the avatar;
retrieving a first user security and privacy policy (USPP) data structure, associated with the operator, specifying security and privacy policies for each affinity group in the plurality of affinity groups, wherein at least two affinity groups have different security and privacy policies specified in the USPP data structure;
applying, based on the classification of the first user into an affinity group, and a security and privacy policy associated with the affinity group as specified in the USPP data structure, an anonymization operation to at least one of static or dynamic characteristics data for the operator which are used to render the avatar, to thereby generate anonymized characteristic data; and
outputting the anonymized characteristic data for rendering the avatar in the virtual environment, wherein the avatar is rendered as a first avatar that has avatar characteristics corresponding to the anonymized versions of the at least one static or dynamic characteristics data,
wherein applying the anonymization operation comprises:
inputting the one of static or dynamic characteristics data for the operator which are used to render the avatar into a trained generator of a trained generative adversarial network (GAN); and
processing, by the generator of the GAN, the one or more static or dynamic characteristics data to generate the anonymized characteristic data, wherein the GAN is trained on training data until a discriminator of the GAN cannot distinguish anonymized characteristic data generated by the generator from actual characteristics of users specified in the training data.
|