US 11,113,730 B1Granted

Parallel data pool processing and intelligent item selection

AssigneeAmazon Technologies, Inc.

InventorsAndrew Douglas Birkett, Andrew J. Bradley, Anirudh Dastidar, Anthony Richard McBryan, Sebastiano Merlino, Fady Mohamed, Sean Daniel Murphy

Life of this patent

Estimating the term…

patentgrep

Loading patent…

Abstract

Systems, methods, and computer-readable media are disclosed for parallel data pool processing and intelligent item selection. In one embodiment, an example method may include determining a first bid request comprising a first user identifier, determining a first set of product identifiers in a first user interaction history of the first user identifier, determining a second bid request comprising a second user identifier, and determining a second set of product identifiers in a second user interaction history of the second user identifier. Example methods may include determining estimated values for one or more product identifiers in the first set of product identifiers and the second set of product identifiers in parallel, and generating respective first and second responses to the first bid request and the second bid request using the estimated values.

Claims20

Dependent claims

A method comprising:

receiving, by one or more computer processors coupled to at least one memory, a first bid request for a content delivery slot, wherein the first bid request comprises a first user identifier for a first user;
determining clickstream data for the first user using the first user identifier;
determining a first set of product identifiers that the first user interacted with using the clickstream data;
deduplicating the first set of product identifiers to remove duplicate product identifiers;
selecting a random subset of the first set of product identifiers, wherein the random subset comprises at most twenty product identifiers;
sending a first evaluation request comprising the random subset to a model evaluator component, wherein the model evaluator component is configured to determine an estimated value for one or more of the random subset, wherein the estimated value comprises a first bid amount for a first product identifier of the one or more of the random subset;
receiving a second bid request comprising a second user identifier for a second user;
determining that the second user interacted with one product identifier;
sending a second evaluation request comprising the one product identifier to the model evaluator component;
determining that a first timer for the for a first response to the first bid request;
determining, based on the model evaluator component, a first set of estimated values for the one or more of the random;
pausing determination of bid amounts for unevaluated product identifiers in the random set based on at least one of a timer value associated with the first timer, a number of unevaluated product identifiers associated with the first bid request, or a number of evaluated product identifiers associated with the first bid request;
determining, during the pausing, an estimated value for the one product identifier;
determining, after the pausing, an estimated value for at least one of unevaluated product identifiers in the random subset;
selecting a highest ranked product identifier from the first set of estimated values and the estimated value for the at least one of unevaluated product identifiers using respective estimated values;
sending the first response to the first bid request using the highest ranked product identifier;
and
sending a second response to the second bid request using the estimated value for the one product identifier.

15 more elements

A method comprising:

determining, by one or more computer processors coupled to at least one memory, a first bid request comprising a first user identifier;
determining a first timer for a first response to the first bid request;
determining a first set of product identifiers in a first user interaction history of the first user identifier;
determining a second bid request comprising a second user identifier;
determining a second set of product identifiers in a second user interaction history of the second user identifier;
determining estimated values for a portion of the first set of product identifiers, wherein the estimated values comprise a first bid amount for a first product identifier of the portion;
pausing determination of bid amounts for unevaluated product identifiers in the first set of product identifiers based on at least one of a timer value associated with the first timer, a number of unevaluated product identifiers associated with the first bid request, or a number of evaluated product identifiers associated with the first bid request;
determining, during the pausing, estimated values for one or more product identifiers in the second set of product identifiers;
determining, after the pausing, an estimated value for at least one of unevaluated product identifiers in the first set of product identifiers;
and
generating respective first and second responses to the first bid request and the second bid request using the estimated values.

8 more elements

A device comprising:

at least one memory that stores computer-executable instructions;
and
at least one processor configured to access the at least one memory and execute the computer-executable instructions to:
determine a first bid request comprising a first user identifier;
determine a first timer for a first response to the first bid request;
determine a first set of product identifiers in a first user interaction history of the first user identifier;
determine a second bid request comprising a second user identifier;
determine a second set of product identifiers in a second user interaction history of the second user identifier;
determine estimated values for a portion of the first set of product identifiers, wherein the estimated values comprise a first bid amount for a first product identifier of the portion;
paus determination of bid amounts for unevaluated product identifiers in the first set of product identifiers based on at least one of a timer value associated with the first timer, a number of unevaluated product identifiers associated with the first bid request, or a number of evaluated product identifiers associated with the first bid request;
determine estimated values for one or more product identifiers the second set of product identifiers;
determine, after the pausing, an estimated value for at least one of unevaluated product identifiers in the first set of product identifiers;
and
generate respective responses to the first bid request and the second bid request using the estimated values.

11 more elements

Dependencies are inferred from the claim text.