Methods and systems for determining provenance and identity of digital advertising requests solicited by publishers and intermediaries representing publishers

A computer implemented method for determining provenance and identity of a programmatic digital advertising request using machine learning to generate a candidate entity from the programmatic digital advertising request for a digital advertisement that is submitted by a candidate entity, wherein the programmatic digital advertisement request is purportedly solicited by one of a publisher and an intermediary representing the publisher that are connected on a digital advertising supply chain, wherein the candidate entity claims to be one of the publisher and the intermediary representing the publisher, wherein the computer implemented method comprises:

• receiving, with a transceiver of a first computing device, a plurality of training messages received from a plurality of candidate entity computing devices;
• wherein each training message has a plurality of training message metrics comprising training serving constraints and training protocols;
• processing each training message from the plurality of training messages by converting at least one training serving constraint and at least one training protocol to a plurality of training message embeddings;
• wherein processing further comprises
• combining, with the processor, the plurality of training message embeddings to a training coordinate tensor;
• identifying, with the processor, a training entity cluster on the training coordinate tensor;
• generating, with the processor using machine learning, based on the training entity cluster, a training entity identity;
• after processing the training message, then generating a machine learning model by training a neural network by storing the training entity identity and the training message on a connected database of the neural network;
• after generating the machine learning model, then receiving, with the transceiver of a first computing device, a first message from a candidate entity computing device associated with the candidate entity;
• wherein the first message comprises the programmatic digital advertising request for the digital advertisement and a plurality of first message metrics;
• wherein a first message metric comprises code defining one of (i) a serving constraint for displaying the digital advertisement and (ii) a protocol of the digital advertisement supply chain;
• wherein the serving constraint has a serving constraint value and the protocol of the digital advertisement supply chain has a protocol value;
• in response to receiving the first message, processing, with a processor of the first computing device, the first message using the machine learning model in the same manner as each of the plurality of training messages, wherein processing the first message comprises;
• generating, using the machine learning model, a first portion of a candidate entity identity;
• and
• wherein the first portion of the candidate entity identity is a first digital identity comprising at least one serving constraint value and at least one protocol value;
• retrieving, from the connected database, at least one training message of the plurality of training messages that is determined by the machine learning model to be associated one of the publisher and the intermediary representing the publisher that the candidate entity purportedly claims to be;
• comparing, with the processor using the machine learning model, at least one of (i) the first message and (ii) the first portion of the candidate entity identity with at least one of (i) the at least one training message and (ii) the training entity identity generated from the at least one training message;
• determining, with the processor, if at least one training message metric of the plurality of training message metrics is not yet identified in the first message;
• if the at least one training message metric is not yet identified in the first message, then processing, with the processor, the first message using the machine learning model, to predict a value, for the at least one training message metric not yet identified in the first message, wherein the predicted value comprises one of (i) the serving constraint value and (ii) the protocol value, for the not yet identified in the first message;
• calculating, with the processor, an instability exhibited by the predicted value relative to the first digital identity, wherein the instability is a measurement of change of a predicted message embedding to change relative to a corresponding embedding of the first digital identity;
• generating, with the processor, a first response to the first message, wherein the first response comprises at least one of (i) sending, with the transceiver, to the candidate entity computing device a second message, and (ii) sending, with the transceiver, to the candidate entity computing device, after a first delay, the second message;
• and
• wherein the second message comprises soliciting a first reply from the candidate entity to the first response wherein the first reply to the first response comprises a third message comprising the at least one training message metric that was not yet identified in the plurality of first message metrics and that exhibited an instability.

A computer implemented method using machine learning for determining provenance and identity of a programmatic digital advertising request and using machine learning to generate a candidate entity identity from the programmatic digital request for a digital advertisement that is submitted by a candidate entity, wherein the programmatic digital advertisement request is purportedly solicited by one of a publisher and an intermediary representing the publisher that are connected on a digital advertisement supply chain, wherein the candidate entity claims to be one of the publisher and the intermediary representing the publisher, wherein the computer implemented method comprises:

• generating, with a processor of a first computing device, a first portion of a candidate entity identity based on at least one candidate entity cluster, wherein generating the portion of the candidate entity identity further comprises:
• receiving, with a transceiver of the first computing device, a first message from a candidate entity computing device associated with the candidate entity, wherein the first message comprises the programmatic digital advertising request for the digital advertisement and a plurality of first message metrics;
• wherein a first message metric comprises code defining one of (i) a serving constraint for displaying the digital advertisement and (ii) a protocol of the digital advertisement supply chain;
• converting, with the processor using machine learning, each first message metric of the plurality of first message metrics to at least one message embedding;
• after converting the plurality of first message metrics to a plurality of first message embeddings, then combining, with the processor, the plurality of first message embeddings to a coordinate tensor;
• identifying, with the processor using machine learning, the at least one candidate entity cluster associated with the plurality of first message embeddings;
• and
• after generating the first portion of the candidate entity identity, then generating, with the processor, a first response to the first message, wherein the first response comprises at least one of (i) sending, with the transceiver, to the candidate entity computing device a second message, and (ii) sending, with the transceiver, to the candidate entity computing device, after a first delay, the second message;
• wherein the second message comprises soliciting a first reply from the candidate entity to the first response wherein the first reply to the first response comprises a third message comprising the at least one publisher message metric not yet identified in the plurality of first message metrics;
• receiving, with the transceiver, from the candidate entity computing device, the first reply to the first response, wherein the first reply comprises the third message comprising the at least one publisher message metric not yet identified in the first message;
• processing, with the processor, the third message in the same manner as the first message to generate a second portion of the candidate entity identity;
• and
• generating the candidate entity identity by combining, with the processor using machine learning, the first portion of the candidate entity identity and the second portion of the candidate entity identity.