Assessment of the accuracy of a learn model machines

摘要

The computer-readable medium on the executable instructions are stored, which performs cause a processor, a computer implemented method for assessing the accuracy of a learn model machines, to carry out the process comprising:the receiving a plurality of requests for a first content element;the assignment of a first part of the received requests for the first content element to a first group and the assigning of the rest of the received requests for the first content element to a second group;the use of a machines learn model for a request which has been allocated to the first group, in order to variants of a plurality of a variant of the first content element to the representation to be selected in response to said query request;the use of a statistical model for a request which has been allocated to the second group, in order to the plurality of variants of a variant of the first content element for the representation in response to said query request; andthe obtaining information about the accuracy of the machines learn model, in that performance data for variants of the first, by means of the machines learn model selected content element with performance data for variants of the first, by means of the statistical model selected content element are compared.