AUTOMATED ALLOCATION OF MEDIA VIA NETWORK

• US 20140278981A1
• Filed: 10/06/2012
• Published: 09/18/2014
• Est. Priority Date: 10/06/2011
• Status: Abandoned Application

• Current Assignee: Kochava, Inc.
• Original Assignee: Kochava, Inc.
• Inventors: Shanahan, James, Mersov, Gueorgui, Prevost, Vince, Pan, Yuanyi, Levine, Alan Steven
• Application Number:
  US14/350,017
• Publication Number:
  US 20140278981A1
• Time in Patent Office: Days
• US Class Current: 705/14.53
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0255 based on user history
• CPC Class Codes: G06Q 30/0275 Auctions
• CPC Class Codes: G06Q 30/08 Auctions

Abstract

There are provided systems and methods for allocation of digital media over communication networks. A bidding platform for digital advertising manages multichannel buying of digital media using impression-level decisioning based on multiple parameters and data sources. When a request for an ad is received from a publisher or other supply side entity, such as an ad exchange or supply side platform, an expected value of the advertisement impression to each advertiser is calculated, in some cases, in real time, on behalf of the advertiser using media-buying rules. An ad having a highest expected value is selected, and the bidding platform responds to the ad request with a bid and the selected ad, and serves the winning ad if bid response wins the publisher side auction, resulting in an ad impression for the winning ad. User interactions with the ad are recorded and leveraged to optimize further the media buying rules.

Claims

1. A system for generating data representing parameters useful for causing a processor to identify content to be displayed on a graphical user interface, the system comprising the same or another processor and stored machine-readable instructions configured to cause the processor to:

parse stored data to identify input patterns associated with a plurality of historical user interactions with displayed content on input/output interfaces, the stored data representing one or more attributes of the plurality of historical user interactions, and the identified input patterns defined in terms of two or more correlated variables;

based at least partly on the identified input patterns, generate data useful in implementing one or more media buying rules useful on a bidding platform for obtaining authorization to provide selected display content data to one or more client systems requesting the same or other selected content for display; and

route the generated data useful in implementing the one or more media buying rules to a bid generation engine associated with the bidding platform.

2. The system of claim 1, wherein the one or more attributes comprises input control commands used in navigation of displayed interactive content.

3. The system of claim 1, wherein the one or more attributes comprises characteristics of the input/output interfaces.

4. The system of claim 1, wherein the one or more attributes comprises characteristics of the content being displayed.

5. The system of claim 1, wherein the one or more attributes comprises characteristics of a user manipulating the input/output interfaces.

6. The system of claim 1, wherein each of the media buying rules comprises:

a conditions component specifying a corresponding one of the identified input patterns; and

a bid component specifying a bid price determined based on a pre-determined cost per action and a predicted conversion rate associated with the corresponding one of the identified input patterns.

7. The system of claim 1, wherein the processor comprises:

a bid optimizer configured to generate the one or more media buying rules based on one or more training data sets stored in a database of data logs and comprising data records of historical interactions with user interfaces.

8. The system of claim 7, wherein the bid optimizer is configured to augment the one or more media buying rules with at least one of first party data, second party data and third party data.

9. The system of claim 7, wherein the bid optimizer comprises:

a feature selector configured to process data logs of historical interactions with user interfaces and to determine features of the historical interactions that have a predictive quality of future interactions.

10. The system of claim 9, wherein the feature selector is configured to:

calculate a test statistic defined between a target variable and at least one context variable recorded in the data logs;

permute the at least one context variable to generate a plurality of different variable permutations;

for each generated permutation of the at least one context variable, calculate a permutation test statistic between the target variable and that permutation of the at least one context variable to generate a plurality of permutation test statistics;

generate a distribution of the permutation of test statistics; and

select or reject the at least one context variable based on a feature of the generated distribution.

11. The system of claim 10, wherein the test statistic comprises mutual information determined between the target variable and the at least one context variable.

12. The system of claim 10, wherein the test statistic comprises a difference in sample means determined between the target variable and the at least one context variable.

13. The system of claim 10, wherein the test statistic comprises a J-measure determined between the target variable and the at least one context variable.

14. The system of claim 10, wherein the test statistic comprises information gain determined between the target variable and the at least one context variable.

15. The system of claim 10, wherein the test statistic comprises a chi-squared statistical measure determined between the target variable and the at least one context variable.

16. The system of claim 10, wherein the feature of the generated distribution comprises a p-value of the determined test statistic.

17. The system of claim 16, wherein the feature selector is configured to select the at least one context variable if it is determined that the p-value is less than a pre-determined threshold.

18. The system of claim 9, wherein the feature selector is configured to partition the data logs into two or more classes of data logs based on a value of the target variable.

19. The system of claim 7, wherein the bid optimizer comprises:

a rules generator configured to generate a plurality of media buying rules from a plurality of variables representing features of historical interactions with user interfaces.

20. The system of claim 19, wherein the rules generator is configured to:

define a plurality of potential media buying rules based on the plurality of variables, each of the plurality of potential media buying rule defined by a corresponding set of values for the plurality of variables;

for each of the plurality of potential media buying rules, compute a corresponding score value used to rank the plurality of potential media buying rules; and

based on the corresponding score value, select a subset of the plurality of potential media buying rules as the plurality of media buying rules.

21-87. -87. (canceled)