Realtime response to network-transferred content requests using statistical prediction models

Patent metadata

Field	Value
Publication number	US-11049022-B2
Application number	US-201715662686-A
Country	US
Kind code	B2
Filing date	Jul 28, 2017
Priority date	Jul 28, 2017
Publication date	Jun 29, 2021
Grant date	Jun 29, 2021

Abstract

Techniques for leveraging existing statistical prediction models are provided. A first statistical prediction model is generated for a content item. An instruction is received to create a clone from the content item. In response to receiving the instruction, the clone is created based on attributes of the content item. A second statistical prediction model that is different than the first statistical prediction model is generated for the clone. In response to receiving a request for content, the clone is identified as relevant to the first request. A similarity between (1) first content of the content item and (2) second content of the clone is determined. If the similarity exceeds a similarity threshold, then the first statistical prediction model is used to generate a prediction of an entity user selection rate associated with the clone. Otherwise, the second statistical prediction model is used to generate the prediction.

First claim

What is claimed is:

1. A system comprising:
one or more processors;
one or more storage media storing instructions which, when executed by the one or more processors, cause:
generating a first statistical prediction model for a particular content item;
receiving an instruction to create a content item clone from the particular content item;
in response to receiving the instruction, creating the content item clone based on attributes of the particular content item;
generating, for the content item clone, a second statistical prediction model that is different than the first statistical prediction model;
receiving a first request for content;
in response to receiving the first request:
identifying the content item clone as relevant to the first request;
determining whether a similarity between (1) first content of the particular content item and (2) second content of the content item clone exceeds a similarity threshold;
if the similarity exceeds the similarity threshold, then using the first statistical prediction model to generate a prediction of an entity user selection rate associated with the content item clone;
if the similarity does not exceed the similarity threshold, then using the second statistical prediction model to generate the prediction of the entity user selection rate;
based on the prediction, selecting the content item clone;
causing the content item clone to be transmitted over a network to be presented on a computing device.

2. The system of claim 1 , wherein:
the particular content item is one of a plurality of content items associated with a particular content campaign;
the instruction is to create a campaign clone of the particular content campaign;
the instructions, when executed by the one or more processors, further cause:
generating, for each content item in the plurality of content items, a statistical prediction model that is different than the first statistical prediction model.

3. The system of claim 1 , wherein the first statistical prediction model is used only if one or more criteria are satisfied, wherein the one or more criteria include a threshold similarity between a first target audience of the particular content item and a second target audience of the content item clone.

4. The system of claim 3 , wherein the threshold similarity includes a first entity selection rate associated with the first target audience of the particular content item being classified as similar to a second entity selection rate associated with the second target audience of the content item clone.

5. The system of claim 1 , wherein the first statistical prediction model is used only if one or more criteria are satisfied, wherein the one or more criteria include a charging model of the particular content item being the same as a charging model of the content item clone.

6. The system of claim 1 , wherein, in response to receiving the first request, using the first statistical prediction model, wherein the instructions, when executed by the one or more processors, further cause:
receiving a second request for content;
in response to receiving the second request:
identifying the content item clone as relevant to the second request;
using the second statistical prediction model to generate a second prediction of the entity user selection rate;
based on the second prediction, selecting the content item clone;
causing the content item clone to be transmitted over the network to be presented on a second computing device.

7. The system of claim 1 , wherein the instructions, when executed by the one or more processors, further cause:
determining a first time indicating when the content item clone became active;
wherein the first statistical prediction model is used only if a difference between a current time and the first time is less than a threshold period of time.

8. The system of claim 7 , wherein the instructions, when executed by the one or more processors, further cause:
receiving an event that indicates the first time and that the content item clone was presented to a user;
in response to determining that the difference between the current time and the first time is less than the threshold period of time:
generating a mapping between an identifier of the content item clone and an identifier of the particular content item;
overwriting, based on the mapping, a copy of the second statistical prediction model with a copy of the first statistical prediction model using the mapping.

9. The system of claim 7 , wherein the instructions, when executed by the one or more processors, further cause:
receiving an event that indicates the first time and that the content item clone was presented to a user;
in response to determining that the difference between the current time and the first time is greater than the threshold period of time:
refrain from generating a mapping between an identifier of the content item clone and an identifier of the particular content item.

10. The system of claim 1 , wherein the instructions, when executed by the one or more processors, further cause:
generating a global statistical prediction model that is based on a plurality of attributes of multiple entities and a plurality of attributes of contexts of requests for content;
generating a plurality of campaign-specific statistical prediction models, each of which is based on a plurality of attributes of content items or a plurality of attributes of users interaction with the content items;
wherein the plurality of campaign-specific statistical prediction models includes the first and second statistical prediction models;
determining to use the first statistical prediction model and the global statistical prediction model in response to a particular request for content;
combining output of the first statistical prediction model and the global statistical prediction model to generate a score that is associated with the content item clone and is used to determine whether to select the content item clone for presentation.

11. A system comprising:
one or more processors;
one or more storage media storing instructions which, when executed by the one or more processors, cause:
generating a statistical prediction model of a particular content item;
receiving an instruction to create a content item clone from the particular content item;
in response to receiving the instruction:
identifying one or more first attribute values of the particular content item;
identifying one or more second attribute values of the content item clone;
performing a comparison of the one or more first attribute values and the one or more second attribute values, wherein the one or more first attribute values and the one or more second attribute values pertain to one or more of the following attributes: target audience, charging model, or content of the respective content items;
based on the comparison, determining whether to associate the statistical prediction model with the content item clone;
if it is determined to associate the statistical prediction model with the content item clone, then using the statistical prediction model to generate a first score for the content item clone in response to a first request for content;
if it is determined to not associate the statistical prediction model with the content item clone, then using a different statistical prediction model to generate a second score for the content item clone in response to a second request for content.

12. The system of claim 11 , wherein the one or more first attribute values are of a first target audience of the particular content item, wherein the one or more second attribute val

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Xu Liqin
• Poon Peter
• Pu Wen
• Karthik Swetha

Classifications

• G06N3/042

  Knowledge-based neural networks; Logical representations of neural networks · CPC title

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06Q30/00

  Commerce · CPC title

• G06Q30/02

  Marketing; Price estimation or determination; Fundraising · CPC title