USING LINEAR AND LOG-LINEAR MODEL COMBINATIONS FOR ESTIMATING PROBABILITIES OF EVENTS

• US 20130275235A1
• Filed: 06/04/2013
• Published: 10/17/2013
• Est. Priority Date: 07/21/2010
• Status: Abandoned Application

• Current Assignee: Excalibur IP, LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Cetin, Ozgur, Manavoglu, Eren, Achan, Kannan, Iyer, Rukmini, Cantu-Paz, Erick
• Application Number:
  US13/909,445
• Publication Number:
  US 20130275235A1
• Time in Patent Office: Days
• US Class Current: 705/14.73
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 30/0241 Advertisements
• CPC Class Codes: G06Q 30/0277 Online advertisement

Abstract

A system for determining predictive models associated with online advertising can include a communications interface, a processor, and a display. The communications interface can be configured to receive a partial dataset. The partial dataset may include user information. The processor can be communicatively coupled to the communications interface and configured to identify the partial dataset. The processor can also be configured to determine a first predictive model corresponding to at least part of the partial dataset and a second predictive model by combining a probability distribution with the first predictive model. The display can be communicatively coupled to the processor and configured to display the second predictive model.

Claims

1. A system for determining predictive models associated with online advertising, comprising:

a communications interface configured to receive a partial dataset, the partial dataset including user information;

a processor, communicatively coupled to the communications interface, configured to;

identify the partial dataset;

determine a first predictive model corresponding to at least part of the partial dataset; and

determine a second predictive model by combining a probability distribution with the first predictive model; and

a display, communicatively coupled to the processor, configured to display the second predictive model.

2. The system of claim 1, wherein the probability distribution is a weighted distribution model.

3. The system of claim 2, wherein the weighted distribution model is a log-linear combination using maximum-entropy weighting.

4. The system of claim 2, wherein the weighted distribution model is a linear combination using uniform average weighting.

5. The system of claim 1, wherein the user information includes user interests.

6. The system of claim 1, where the user information includes user demographics.

7. The system of claim 1, where the user information includes user web browsing behaviors.

8. The system of claim 7, where the web browsing behaviors include historical data regarding one or more of user queries, click-throughs, or other measurable web browsing events.

9. The system of claim 1, wherein the second predictive model (p(c|x)) uses the formula:

10. The system of claim 1, wherein the second predictive model (p(c|x)) uses the formula:

11. The system of claim 1, wherein the determination of the first predictive model includes training a predictive model using flat weighting.

12. The system of claim 1, wherein the processor is further configured to partition the user information into categories.

13. The system of claim 12, wherein the categories include one or more of user interests, user demographics, and user web browsing behaviors.

14. A method for determining predictive models associated with online advertising, comprising:

receiving, at a communications interface, a partial dataset including user information;

identifying the partial dataset by a processor communicatively coupled to the communications interface;

determining, by the processor, a first predictive model corresponding to at least part of the partial dataset; and

determining, by the processor, a second predictive model by combining a probability distribution with the first predictive model.

15. The method of claim 14, further comprising displaying the second predictive model at a display communicatively coupled to the processor.

16. The method of claim 14, wherein the probability distribution is a weighted distribution model or a log-linear combination using maximum-entropy weighting, and wherein the weighted distribution model is a linear combination using uniform average weighting.

17. The method of claim 14, wherein the user information includes one or more of user interests, user demographics, and user web browsing behaviors, and wherein the web browsing behaviors include historical data regarding one or more of user queries, click-throughs, or other measurable web browsing events.

18. The method of claim 14, wherein the second predictive model (p(c|x)) uses the formula:

19. The method of claim 14, wherein the processor is further configured to partition the user information into categories, and wherein the categories include one or more of user interests, user demographics, and user web browsing behaviors.

20. A non-transitory computer readable medium for determining predictive models associated with online advertising, comprising:

instructions executable by a processor to receive a partial dataset, the partial dataset including user information;

instructions executable by a processor to identify the partial dataset;

instructions executable by a processor to determine a first predictive model corresponding to at least part of the partial dataset; and

instructions executable by a processor to determine a second predictive model by combining a probability distribution with the first predictive model.