Modifying advertisement scores based on advertisement response probabilities

• US 7,370,002 B2
• Filed: 06/05/2002
• Issued: 05/06/2008
• Est. Priority Date: 06/05/2002
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Luo, Martin, Ali, Mahbubul Alam, Heckerman, David E., Shani, Guy
• Primary Examiner(s): Hafiz; Tariq R.
• Assistant Examiner(s): PATS, JUSTIN
• Application Number:
  US10/163,056
• Publication Number:
  US 20030229531A1
• Time in Patent Office: 2,162 Days
• Field of Search: 705/10, 705/14
• US Class Current: 705/14.41
• 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 modifying advertisement scores based on advertisement response probabilities and targeting advertisements to users based on a probability that a user will respond to possible advertisements, comprising the following:

receiving a request to display a page of web content to a recipient, the page of web content including an advertisement portion which is dynamically variable to include advertising content selected based at least in part on a probability that the recipient will respond to a selected advertisement;

accessing a plurality of possible advertisements available for selection to be inserted into the advertisement portion;

receiving a base score for each of the plurality of possible advertisements;

receiving a probability prediction model for each of the plurality of possible advertisements, such that each of the plurality of possible advertisements has a different corresponding probability prediction model, and wherein the probability prediction model for each of the plurality of possible advertisements includes a plurality of response probabilities;

for each of the plurality of possible advertisements, selecting one of the plurality of response probabilities included in the corresponding probability prediction model;

calculating an average response probability, wherein the average response probability is based on each of the plurality of response probabilities from each of the different probability prediction models; and

for the plurality of possible advertisements, altering the base score of the corresponding possible advertisements based on an amount by which the selected response probability for each respective possible advertisement deviates from the average response probability, wherein altering the base score for each possible advertisement comprises;

applying a first multiplicative factor to all base scores corresponding to the possible advertisements having a selected response probability greater than the average response probability plus a predetermined number of standard deviations of the selected response probabilities, wherein applying the first multiplicative factor increases the corresponding scores; and

applying a second multiplicative factor to all base scores corresponding to the possible advertisements having a selected response probability less than the average response probability minus the predetermined number of standard deviations of the selected response probabilities, wherein the second multiplicative factor decreases the corresponding scores.

2. The method as recited in claim 1, wherein accessing a plurality of possible advertisements includes accessing the plurality of possible advertisements from an advertisement database.

3. The method as recited in claim 1, wherein accessing a plurality of possible advertisements includes accessing the plurality of possible advertisements from an advertisement pipeline.

4. The method as recited in claim 1, wherein the plurality of different probability prediction models include a different decision tree for each of the plurality of possible advertisements.

5. The method as recited in claim 4, wherein each decision tree includes the plurality of response probabilities only at leaf nodes.

6. The method as recited in claim 4, wherein each decision tree includes a base node, a plurality of intermediate nodes, and a plurality of leaf nodes, wherein each of the nodes corresponds to a different recipient attribute.

7. The method as recited in claim 6, wherein only the leaf nodes include response probabilities, and wherein each leaf node contains at least one response probability.

8. The method as recited in claim 4, wherein selecting one of the plurality or response probabilities includes traversing the decision tree corresponding to each possible advertisement to reach a leaf node that contains the selected response probability.

9. The method as recited in claim 1, wherein the plurality of different probability prediction models include at least one naï

ve Bayes predictive model.

10. The method of claim 1, wherein the plurality of different probability prediction models include at least one logical regression predictive model.

11. The method of claim 1, wherein selecting one of the plurality of response probabilities for each of the plurality of possible advertisements includes selecting the one response probability corresponding to an attribute value associated with the recipient.

12. The method of claim 1, wherein each of the plurality of response probabilities reflects a probability a user will click on a respective advertisement.

13. The method of claim 1, wherein each of the plurality of response probabilities reflects a probability a user will buy on a product advertised in a respective advertisement.

14. The method of claim 1, wherein altering the corresponding base score for the plurality of possible advertisement comprises increasing base scores of advertisements having a response probability greater than the average response probability.

15. The method of claim 1, wherein the average is a weighted average based on a percentage of times a leaf node is used to select a response probability.

16. The method of claim 1, wherein altering the corresponding base score for the plurality of possible advertisement comprises decreasing base scores of advertisements having a response probability less than the average response probability.

17. The method of claim 1, wherein altering the base score for the possible advertisements further comprises altering the base score for possible advertisements between the average response probability plus the predetermined number of standard deviations of the selected response probabilities and the average response probability minus the predetermined number of standard deviations of the selected response probabilities by an amount proportional to the deviation of the corresponding selected response value from the average response value.

18. The method of claim 1, wherein the predetermined number of standard deviations is one.

19. The method of claim 1, wherein the predetermined number of standard deviations is two.

20. The method of claim 1, wherein the standard deviation is a weighted standard deviation based on a weighted average response probability.

21. The method of claim 1, further comprising:

selecting one of the possible advertisements, the selected one of the possible advertisements having the highest altered score; and

inserting the selected one of the possible advertisements into the advertisement portion of the page of web content.

22. The method of claim 1, wherein receiving the base score for each of the possible advertisements includes accessing a base score stored with its corresponding possible advertisement.

23. The method of claim 1, wherein receiving a probability prediction model for each of the plurality of possible advertisements includes accessing a probability prediction model stored with its corresponding possible advertisement.

24. A computer program product for use in a network environment that includes at least an advertisement computer system, the computer program product comprising one or more computer-readable media having stored thereon computer executable instructions for performing the method of claim 1.

25. A network system comprising:

an advertisement computer system, the advertising computer system having access to computer executable instructions that, when executed by the advertisement computer system, perform the method of claim 1.

26. A network system as recited in claim 25, wherein the computer executable instructions further cause the advertisement computer system to select one of the plurality of possible advertisements to be presented to the recipient.

27. A network system as recited in claim 25, further comprising a user computer system having access to computer executable instructions that, when executed by the user computer system, cause the user computer system to display the selected one of the plurality of possible advertisements to the recipient.