Optimization of real-time probabilistic model evaluation for online advertising

Patent metadata

Field	Value
Publication number	US-10565622-B1
Application number	US-201514667521-A
Country	US
Kind code	B1
Filing date	Mar 24, 2015
Priority date	Mar 24, 2015
Publication date	Feb 18, 2020
Grant date	Feb 18, 2020

Abstract

Systems, methods, and computer-readable media are disclosed for optimization of real-time probabilistic model evaluation for online advertising. In one embodiment, a system may receive a bid request for an ad slot, and may receive cached user information with a first ad and a first ad score, and a second ad and a second ad score. The system may generate a first estimated probability of conversion associated with presentation of the first ad to the user, and may generate a second estimated probability of conversion associated with presentation of the second ad to the user based at least in part on the context information, the second ad identifier, and the second ad score. The system may select either the first ad or the second ad for which to generate a bid amount based at least in part on the first estimated probability of conversion and the second estimated probability of conversion.

First claim

What is claimed is:

1. A method comprising:
receiving, by a server comprising one or more computer processors, a bid request for an ad slot from an ad exchange server, the bid request comprising website context information of the ad slot and an account profile identifier indicative of a user account for which the ad slot is available;
sending, by the server, a cached user information request to a remote server, the cached user information request comprising a request for pre-calculated account profile information for the user that is stored in a cache, the pre-calculated account profile information comprising pre-matched ads eligible for presentation to the user;
receiving, by the server and in response to the cached user information request, cached user information comprising a first eligible ad identifier for a first eligible ad, a first ad score indicative of a relationship between targeting criteria of the first eligible ad and the account profile identifier, a second eligible ad identifier for a second eligible ad, and a second ad score indicative of a relationship between targeting criteria of the second eligible ad and the account profile identifier, wherein the first ad score is a first numerical representation of the relationship between the targeting criteria and the account profile identifier for the first eligible ad, and the second ad score is a second numerical representation of the relationship between the targeting criteria and the account profile identifier for the second eligible ad, and wherein the cached user information comprises a pre-ordered ranking of the first eligible ad and the second eligible ad based on the first ad score, the second ad score, and historical data indicative of a presentation history of the first eligible ad and the second eligible ad to the user;
generating, by the server, a first estimated probability of conversion for presentation of the first eligible ad to the user based on the website context information, the first eligible ad identifier, and the first ad score, the first estimated probability of conversion indicative of a likelihood the first eligible ad will be interacted with;
generating, by the server, a second estimated probability of conversion for presentation of the second eligible ad to the user based on the website context information, the second eligible ad identifier, and the second ad score, the second estimated probability of conversion indicative of a likelihood the second eligible ad will be interacted with;
determining, by the server, that the first estimated probability of conversion is greater than the second estimated probability of conversion;
determining, by the server, a comparison between a rate of ad impression delivery associated with the first eligible ad and a threshold rate of ad impression delivery; and
causing the second eligible ad to be placed at the ad slot for presentation to the user instead of the first eligible ad based at least in part on the comparison.

2. The method of claim 1 , further comprising:
extracting, by the server, the website context information and the account profile identifier from the bid request to generate extracted information; and
sending, by the server, the extracted information to the remote server with the cached user information request.

3. The method of claim 1 , further comprising sending, by the server, a set of candidate ads to the remote server, the set of candidate ads comprising available ad inventory from which ads can be selected for presentation to users, wherein the first eligible ad and the second eligible ad are selected from the set of candidate ads.

4. A method comprising:
receiving, by one or more servers comprising one or more computer processors, a bid request for an ad slot, the bid request comprising available ad slot information for an ad slot for presentation to a user;
receiving, by the one or more servers, cached user information from a remote server, the cached user information comprising a first ad identifier for a first ad, a first ad score indicative of a relationship between the first ad and the user, a second ad identifier for a second ad, and a second ad score indicative of a relationship between the second ad and the user, wherein the cached user information comprises a pre-ordered ranking of the first ad and the second ad based on the first ad score, the second ad score, and historical data indicative of a presentation history of the first ad and the second ad to the user;
generating, by the one or more servers, a first estimated probability of conversion associated with presentation of the first ad to the user based at least in part on the available ad slot information, the first ad identifier, and the first ad score;
generating, by the one or more servers, a second estimated probability of conversion associated with presentation of the second ad to the user based at least in part on the available ad slot information, the second ad identifier, and the second ad score;
selecting, by the one or more servers, the first ad for which to generate a bid amount based at least in part on determining that the first estimated probability of conversion is greater than the second estimated probability of conversion;
determining, by the one or more servers, a comparison between a rate of ad impression delivery associated with the first ad and a threshold rate of ad impression delivery; and
causing the second ad to be placed at the ad slot for presentation to the user instead of the first ad based at least in part on the comparison.

5. The method of claim 4 , further comprising:
generating, by the one or more servers, a bid amount for the second ad based at least in part on the second estimated probability of conversion; and
submitting, by the one or more servers, the bid amount in response to the bid request.

6. The method of claim 4 , wherein the available ad slot information comprises context information representing context for the ad slot and a user identifier associated with the user.

7. The method of claim 6 , further comprising:
extracting, by the one or more servers, the context information and the user identifier from the available ad slot information to generate extracted information; and
sending, by the one or more servers, a cached user information request to the remote server, the cached user information request comprising a request for user information of the user and the extracted information.

8. The method of claim 7 , further comprising sending, by one or more servers, a set of candidate ads to the remote server, the set of candidate ads comprising available ad inventory from which ads can be selected for presentation to the user, wherein the first ad and the second ad are selected from the set of candidate ads.

9. The method of claim 8 , wherein the cached user information comprises a set of eligible ads selected from the set of candidate ads, the set of eligible ads comprising the first ad and the second ad, the set of eligible ads indicative of ads with targeting criteria for which the user meets a threshold level of the targeting criteria.

10. The method of claim 4 , further comprising:
determining, by the one or more servers, that the first ad was previously presented to the user; and
selecting, by the one or more servers, the second ad for which to generate the bid amount.

11. The method of claim 4 , further comprising:
determining, by the one or more servers, a conversion event indicated by a user interaction with the second ad; and
sending, by the one or more servers, a user information update event notification to the remote server indicating that the user information associated with the user is to be updated based at

Assignees

• Amazon Tech Inc

Inventors

• Burstein Jonathan Lee
• Huang Jim

Classifications

• G06Q30/0275

  Auctions · CPC title

• G06Q30/0266 Primary

  based on the position of the vehicle · CPC title