Traffic predictor for network-accessible information modules

• US 7,987,261 B2
• Filed: 07/31/2007
• Issued: 07/26/2011
• Est. Priority Date: 07/31/2007
• Status: Active Grant

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Gamble, Nathan E.
• Primary Examiner(s): WASEL, MOHAMED A
• Application Number:
  US11/831,075
• Publication Number:
  US 20090037421A1
• Time in Patent Office: 1,456 Days
• Field of Search: 709/223, 709/224, 709/226
• US Class Current: 709/224
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Methods, apparatuses and systems directed to predicting the performance of an information package or module presented on a network addressable resource, such as a web page. A particular implementation relies on regression models that utilize statistical measures of user interest in terms, concepts and other subject matter as revealed in on-line search activity of a pool of users. A model receives as inputs a plurality of attribute values corresponding to an information package or module (including one or more statistical measures of interest) and outputs an estimated click-thru rate.

Claims

1. A method for estimating the performance of an information package, comprising operations of:

receiving one or more values of respective attributes of an information package;

obtaining for at least one of the one or more attributes, a buzz value characterizing the level of interest in the attribute based on analysis of online user search activity, wherein the obtaining operation further comprises;

accessing a search system maintaining buzz values for a plurality of terms, topics or categories;

mapping one or more of the attributes of the information package to respective terms, topics or categories; and

estimating a future performance of the information package by providing one or more attributes and one or more buzz values as input parameters to a regression model to yield a numerical value characterizing the estimated performance of the information package, wherein the regression model is based on historical performance of a set of previous information packages relative to the one or more attributes, and provides the coefficients for the input parameters to the estimated performance of the information package and wherein each operation of the method is executed by one or more processors.

2. The method of claim 1 wherein the estimated performance comprises a click-thru rate of the information package.

3. The method of claim 2 further comprising determining, by the one or more computing devices, an expected number of clicks based on said click-thru rate and an estimated number of impressions.

4. The method of claim 3 wherein one or more of the attributes comprise temporal information characterizing the time parameters according to which the information package will be presented;

and wherein the method further comprises accessing, by the one or more computing devices, a data store of page impression information to estimate a number of impressions for the information package.

5. The method of claim 1 wherein one or more of the attributes comprise temporal information characterizing the time parameters according to which the information package will be presented;

and wherein the method further comprises accessing, by the one or more computing devices, a data store of page impression information to estimate a number of impressions for the information package.

6. The method of claim 1 further comprising multiplying, by the one or more computing devices, each value of the one or more attributes by the corresponding coefficient provided by the regression model to yield an adjusted value, and summing the adjusted values for each attribute to yield the expected performance of the information package.

7. The method of claim 1 further comprising constructing, by the one or more computing devices, the regression model and selecting, by the one or more computing devices, one or more attributes for analysis based on an assessment of which attributes are statistically significant predictors of the performance of information packages.

8. The method of claim 1 wherein the information package is a network accessible information package.

9. The method of claim 1 further comprising optimizing, by the one or more computing devices, one or more attributes of the information package based on the estimated performance.

10. An apparatus comprising:

one or more network interfaces;

a memory;

one or more processors; and

a performance predictor module, comprising computer-readable instructions physically stored in the memory, operable to cause the one or more processors to perform the following operations;

receive one or more values of respective attributes of an information package;

obtain, for at least one of the one or more attributes, a buzz value characterizing the level of interest in the attribute based on analysis of on-line user search activity, wherein this operation further includes the operations;

access a search system maintaining buzz values for a plurality of terms, topics or categories; and

map one or more of the attributes of the information package to respective terms, topics or categories; and

compute an estimated future performance of the information package by applying one or more of the attributes and buzz values to a regression model, wherein the regression model is based on historical performance of a set of previous information packages relative to the one or more attributes.

11. The apparatus of claim 10 wherein the estimated performance comprises a click-thru rate of the information package.

12. The apparatus of claim 11 wherein the performance predictor module further comprises instructions operative to cause the one or more processors to determine an expected number of clicks based on the estimated click-thru rate and an estimated number of impressions.

13. The apparatus of claim 12 wherein one or more of the attributes comprise temporal information characterizing the time parameters according to which the information package will be presented;

and wherein the performance predictor module further comprises instructions operative to cause the one or more processors to access a data store of page impression information to estimate a number of impressions for the information package.

14. The apparatus of claim 10 wherein one or more of the attributes comprise temporal information characterizing the time parameters according to which the information package will be presented;

and wherein the performance predictor module further comprises instructions operative to cause the one or more processors to access a data store of page impression information to estimate a number of impressions for the information package.

15. The apparatus of claim 10 wherein the performance predictor module further comprises instructions operative to cause the one or more processors to multiply each value of the one or more attributes by the corresponding coefficient provided by the regression model to yield an adjusted value, and sum the adjusted values for each attribute to yield the expected performance of the information package.

16. The apparatus of claim 10 wherein the performance predictor module further comprises instructions operative to cause the one or more processors to construct the regression model and select one or more attributes for analysis based on an assessment of which attributes are statistically significant predictors of the performance of information packages.

17. The apparatus of claim 10 wherein the information package is a network accessible information package.

18. The apparatus of claim 10 wherein the performance predictor module further comprises instructions operative to cause the one or more processors to optimize one or more attributes of the information package based on the estimated performance.

19. A method comprising operations of:

defining, using a computing device, one or more values of respective attributes of an information package;

obtaining, for at least one of the one or more attributes, a buzz value characterizing the level of interest in the attribute based on analysis of on-line user search activity, wherein the obtaining operation further comprises mapping one or more of the attributes of the information package to respective terms, topics, or categories having buzz values; and

providing one or more attributes and one or more buzz values as input parameters to a regression model to yield an estimated future click-thru rate for the information package, wherein the regression model is based on historical click-thru rate performance of a set of previous information packages relative to the one or more attributes, and provides the coefficients for the input parameters to the estimated click-thru rate for the information package; and

optimizing the information package by iteratively redefining one or more values of the respective attributes of the information package and applying the redefined attribute values and corresponding buzz values to the regression model until a desired estimated future click-thru rate is achieved.