Modifying advertisement scores based on advertisement response probabilities

• US 20030229531A1
• Filed: 06/05/2002
• Published: 12/11/2003
• Est. Priority Date: 06/05/2002
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Heckerman, David E., Shani, Guy, Luo, Martin, Ali, Mahbubul Alam
• Granted Patent: US 7,370,002 B2
• Time in Patent Office: Days
• US Class Current: 705/10
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0204 Market segmentation
• CPC Class Codes: G06Q 30/0242 Determining effectiveness of advertisements
• CPC Class Codes: G06Q 30/0254 based on statistics
• CPC Class Codes: G06Q 30/0269 based on user profile or attribute
• CPC Class Codes: G06Q 30/0277 Online advertisement

Abstract

Advertisement response probabilities are utilized to alter advertisement scores. A plurality of possible advertisements is accessed from, for example, an advertisement database or advertisement pipeline. A response probability for each advertisement is determined. A response probability may be a probability that a user will “click,” or otherwise select an advertisement. Advertisements may be associated with probabilistic prediction models that take advertisement recipient attribute values as inputs and provide a probability distribution as output. A score associated with each of the possible advertisements is altered based on the response probability for each of the advertisements. Statistical prediction is used to determine how scores are to be altered. Advertisements with response probabilities less than a mean probability may have associated scores decreased. Conversely, advertisements with response probabilities greater than a mean probability may have associated scores increased.

Claims

1. In a network environment that includes at least an advertisement computer system, a method for targeting online advertisements to users based on the probability that a user will respond to possible advertisements, comprising the following:

an act of accessing a plurality of possible advertisements;

an act of determining a response probability for each advertisement in the plurality of possible advertisements, the response probability representing a probability that a user will respond to an advertisement; and

an act of altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements, the score being used to indicate whether or not to present the advertisement to the user.

2. The method as recited in claim 1, wherein the act of accessing a plurality of possible advertisements comprises the following:

an act of accessing a plurality of possible advertisements from an advertisement database.

3. The method as recited in claim 1, wherein the act of accessing a plurality of possible advertisements comprises the following:

an act of accessing a plurality of possible advertisements from an advertisement pipeline.

4. The method as recited in claim 1, wherein the act of determining a response probability for each advertisement in the plurality of possible advertisements c omprises the following:

an act of using a probabilistic prediction model to determine a response probability for each advertisement in the plurality of possible advertisements.

5. The method as recited in claim 4, wherein the act of using a probabilistic prediction model to determine a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of using a decision tree to determine a response probability for an advertisement.

6. The method as recited in claim 5, wherein the act of using a decision tree to determine a response probability for an advertisement comprises the following:

an act of traversing the decision tree to reach a leaf node that contains a response probability.

7. The method as recited in claim 4, wherein the act of using a probabilistic prediction model to determine a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of using a naive Bayes predictive model to determine a response probability for each advertisement in the plurality of possible advertisements.

8. The method as recited in claim 4, wherein the act of using a probabilistic prediction model to determine a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of using a logistical regression predictive model to determine a response probability for each advertisement in the plurality of possible advertisements.

9. The method as recited in claim 1, wherein the act of determining a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of using recipient attribute values to determine a response probability for each advertisement in the plurality of possible advertisements.

10. The method as recited in claim 1, wherein the act of determining a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of using a probability distribution to determine a response probability for each advertisement in the plurality of possible advertisements.

11. The method as recited in claim 1, wherein the act of determining a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of determining a probability that a user will click on an advertisement.

12. The method as recited in claim 1, wherein the act of determining a response probability for each advertisement in the plurality of possible advertisements comprises the following:

an act of determining a probability that a user will buy an advertised product.

13. The method as recited in claim 1, wherein the act of altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements comprises the following:

an act of increasing scores associated with advertisements that have a response probability greater than a designated value.

14. The method as recited in claim 13, wherein the act of increasing scores associated with advertisements that have a response probability greater than a designated value comprises the following:

an act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution.

15. The method as recited in claim 14, wherein the act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of increasing scores that have a response probability greater than a weighted average of the response probabilities included in a probability distribution.

16. The method as recited in claim 14, wherein the act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in all the leaf nodes of decision trees associated with each of the possible advertisements.

17. The method as recited in claim 14, wherein the act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of increasing a first set of scores associated with advertisements that have a response probability greater than the mean value plus a specified number of standard deviations of the response probabilities in a probability distribution, by multiplying the scores by a first multiplicative factor.

18. The method as recited in claim 17, wherein the act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of increasing a second set of scores associated with advertisements that have a response probability less than the mean value plus a specified number of standard deviations of the response probabilities in the probability distribution, by multiplying the scores by a second multiplicative factor.

19. The method as recited in claim 18, wherein the first multiplicative factor has a magnitude greater than the second multiplicative factor.

20. The method as recited in claim 14, wherein the act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution plus a specified number of weighted standard deviations of the response probabilities included in the probability distribution.

21. The method as recited in claim 14, wherein the act of increasing scores associated with advertisements that have a response probability greater than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of increasing scores that have a response probability greater than a mean value of the response probabilities included in a probability distribution that was output from a naive Bayes predictive model.

22. The method as recited in claim 1, wherein the act of altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements comprises the following:

an act of decreasing scores associated with advertisements that have a response probability less than a designated value.

23. The method as recited in claim 22, wherein the act of decreasing scores associated with advertisements that have a response probability less than a designated value comprises the following:

an act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution.

24. The method as recited in claim 23, wherein the act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of decreasing scores that have a response probability greater than a weighted average of the response probabilities included in a probability distribution.

25. The method as recited in claim 23, wherein the act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in all the leaf nodes of decision trees associated with each of the possible advertisements.

26. The method as recited in claim 23, wherein the act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of decreasing a first set of scores associated with advertisements that have a response probability less than the mean value minus a specified number of standard deviations of the response probabilities in a probability distribution, by multiplying the scores by a third multiplicative factor.

27. The method as recited in claim 26, wherein the act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of decreasing a second set of scores associated with advertisements that have a response probability greater than the mean value minus a specified number of standard deviations of the response probabilities in a probability distribution, by multiplying the scores by a fourth multiplicative factor.

28. The method as recited in claim 27, wherein the third multiplicative factor has a magnitude less than the fourth multiplicative factor.

29. The method as recited in claim 23, wherein the act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution plus a specified number of weighted standard deviations of the response probabilities included in the probability distribution.

30. The method as recited in claim 23, wherein the act of decreasing scores associated with advertisements that have a response probability less than the mean value of the response probabilities included in a probability distribution comprises the following:

an act of decreasing scores that have a response probability less than a mean value of the response probabilities included in a probability distribution that was output from a logistical regression predictive model.

31. The method as recited in claim 1, wherein the act of altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements comprises the following:

an act of using statistical prediction to alter a score associated with each of the possible advertisements.

32. The method as recited in claim 1, wherein the act of altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements comprises the following:

an act of altering a score associated with an advertisement based on the probability that a user will click on the advertisement.

33. The method as recited in claim 1, further comprising:

an act of associating an altered scored with each of the possible advertisements.

34. The method as recited in claim 33, further comprising:

an act of outputting each of the possible advertisements.

35. The method as recited in claim 34, wherein an act of outputting each of the possible advertisements comprises the following:

an act of outputting each of the possible advertisements to an advertisement database.

36. The method as recited in claim 34, wherein an act of outputting each of the possible advertisements comprises the following:

an act of outputting each of the possible advertisements to an advertisement pipeline.

37. In a network environment that includes at least an advertisement computer system, a method for targeting online advertisements to users based on the probability that a user will select possible advertisements, comprising the following:

a step for identifying a response probability for each advertisement in a plurality of possible advertisements; and

an act of altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements, the score being used to indicate whether or not to present the advertisement to the user.

38. A computer program product for use in a network environment that includes at least an advertisement computer system, the computer program product for implementing a method for targeting online advertisements to users based on the probability that a user will respond to possible advertisements, the computer program product comprising one or more computer-readable media having stored thereon the following:

computer-executable instructions for accessing a plurality of possible advertisements;

computer-executable instructions for determining a response probability for each advertisement in the plurality of possible advertisements, the response probability representing a probability that a user will respond to an advertisement;

and

computer-executable instructions for altering a score associated with each of the possible advertisements based on the response probability for each of the possible advertisements, the score being used to indicate whether or not to present the advertisement to the user.

39. The computer program product as recited claim 38, wherein the one or more computer-readable media include physical storage media.

40. The computer program product as recited claim 38, wherein the one or more computer-readable media include system memory.

41. A network system, comprising:

a user computer system configured to present advertisements to a user; and

an advertisement computer system that is network connectable to the user computer system and that is configured to access a plurality of advertisements, to determine a response probability for each advertisement in the plurality of advertisements, and to alter a score for each advertisement based on the response probability for each advertisement.

42. In a network environment that includes at least a Web content computer system, a method for targeting Web content to users based on the probability that a user will respond to possible Web content, comprising the following:

an act of accessing a plurality of possible portions of Web content;

an act of determining a response probability for each portion of Web content in the plurality of possible portions of Web content, the response probability representing a probability that a user will respond to a portion of Web content; and

an act of altering a score associated with each of the possible portions of Web content based on the response probability for each of the possible portions of Web content, the score being used to indicate whether or not to present the portion of Web content to the user.

43. The method as recited in claim 42, wherein the act of accessing a plurality of possible portions of Web content comprises the following:

an act of accessing a plurality of Web pages.

44. The method as recited in claim 42, wherein the act of determining a response probability for each portion of Web content in the plurality of possible portions of Web content comprises the following:

an act of using a probabilistic prediction model to determine a response probability for each portion of Web content in the plurality of possible portions of Web content.