Managing online content based on its predicted popularity

1. A method, comprising:

determining a historical popularity value for a user-selectable online content from historical data describing user accesses to the user-selectable online content over a selected period and a determined age of the user-selectable online content;

ascertaining from the historically popularity value a predicted popularity value describing future popularity of the user-selectable online content at a future time after the selected period; and

managing a web site from which user-selectable online content is accessible based on the predicted popularity value;

wherein the determining, the ascertaining, and the managing are performed by a computer.

2. The method of claim 1 , wherein the determining comprises determining a count of accesses to the user-selectable online content, normalizing the count by a normalization factor to determine a normalized age of the user-selectable online content, and determining the selected period from the normalized age.

3. The method of claim 1 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a linear model that relates a logarithmic transformation of the predicted popularity value to a logarithmic transformation of the historical popularity value.

4. The method of claim 1 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a linear regression model of historical popularity values of other user-selectable online contents accessible from the web site.

5. The method of claim 4 , wherein the ascertaining comprises ascertaining the value {circumflex over (N)} s (t i ,t r ) of the predicted popularity at an age t r for each content s by evaluating

{circumflex over (N)} s ( t i ,t r )=exp[ln N s ( t i )+β 0 ( t i )+σ 0 2 /2]

wherein t i is an age of the content s when the prediction was made, N s (t i ) is a value of popularity of the content s at the age t i , β 0 (t i ) is a maximum likelihood parameter estimator for an intercept of the linear regression model with slope 1 at the age t i , and σ 0 2 is a value of variance of residuals of the regression model on a logarithmic scale.

6. The method of claim 1 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a constant scaling model of historical popularity values of other user-selectable online contents accessible from the web site.

7. The method of claim 6 , wherein the ascertaining comprises ascertaining the value {circumflex over (N)} s (t i ,t r ) of the predicted popularity at an age t r for each content s by evaluating

{circumflex over (N)} s ( t i ,t r )=α( t i ,t r ) N s ( t i )

wherein t i is an age of the content s when the prediction was made, N s (t i ) is a value of popularity of the content s at the age t i , and α(t i ,t r ) has a parameter value that depends on t i and t r .

8. The method of claim 7 , wherein α(t i ,t r ) is calculated by evaluating

α

⁡

(

t

i

,

t

r

)

=

∑

c

⁢

N

c

⁡

(

t

i

)

N

c

⁡

(

t

r

)

∑

c

⁢

[

N

c

⁡

(

t

i

)

N

c

⁡

(

t

r

)

]

2

wherein N c (t i ) are popularity values of content in a set c of user-selectable online contents accessible from the web site at the age t i , and N c (t r ) are popularity values of content in the set c of user-selectable online contents accessible from the web site at the age t r .

9. The method of claim 1 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a growth profile model of historical popularity values of other user-selectable online contents accessible from the web site.

10. The method of claim 9 , wherein the ascertaining comprises ascertaining the value {circumflex over (N)} s (t i ,t r ) of the predicted popularity at an age t r for each content s by evaluating

N

^

s

⁡

(

t

i

,

t

r

)

-

N

s

⁡

(

t

i

)

P

⁡

(

t

i

,

t

r

)

wherein t i is an age of the content s when the prediction was made, N s (t i ) is a value of popularity of the content s at the age t i , and P(t i ,t r ) is an average of relative popularities for a set of contents accessible from the web site at the ages t i and t r .

11. The method of claim 1 , further comprising performing the determining and the ascertaining for each of multiple user-selected online contents, and the managing comprises determining a prioritization order of the user-selected contents in respective prioritized positions on the web page based on the ascertained predicted popularity values.

12. Apparatus, comprising:

a computer-readable medium storing computer-readable instructions; and

a data processor coupled to the computer-readable medium, operable to execute the instructions, and based at least in part on the execution of the instructions operable to perform operations comprising

determining a historical popularity value for a user-selectable online content from historical data describing user accesses to the user-selectable online content over a selected period and a determined age of the user-selectable online content,

ascertaining from the historically popularity value a predicted popularity value describing future popularity of the user-selectable online content at a future time after the selected period, and

managing a web site from which user-selectable online content is accessible based on the predicted popularity value.

13. The apparatus of claim 12 , wherein in the ascertaining the data processor is operable to perform operations comprising ascertaining the predicted popularity value based on a linear model that relates a logarithmic transformation of the predicted popularity value to a logarithmic transformation of the historical popularity value.

14. The apparatus of claim 12 , wherein in the ascertaining the data processor is operable to perform operations comprising ascertaining the predicted popularity value based on a linear regression model of historical popularity values of other user-selectable online contents accessible from the web site.

15. The apparatus of claim 12 , wherein in the ascertaining the data processor is operable to perform operations comprising ascertaining the predicted popularity value based on a constant scaling model of historical popularity values of other user-selectable online contents accessible from the web site.

16. The apparatus of claim 12 , wherein in the ascertaining the data processor is operable to perform operations comprising ascertaining the predicted popularity value based on a growth profile model of historical popularity values of other user-selectable online contents accessible from the web site.

17. At least one non-transitory computer-readable medium having computer-readable program code embodied therein, the computer-readable program code adapted to be executed by a computer to implement a method comprising:

determining a historical popularity value for a user-selectable online content from historical data describing user accesses to the user-selectable online content over a selected period and a determined age of the user-selectable online content;

ascertaining from the historically popularity value a predicted popularity value describing future popularity of the user-selectable online content at a future time after the selected period; and

managing a web site from which user-selectable online content is accessible based on the predicted popularity value.

18. The at least one computer-readable medium of claim 17 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a linear model that relates a logarithmic transformation of the predicted popularity value to a logarithmic transformation of the historical popularity value.

19. The at least one computer-readable medium of claim 17 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a linear regression model of historical popularity values of other user-selectable online contents accessible from the web site.

20. The at least one computer-readable medium of claim 17 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a constant scaling model of historical popularity values of other user-selectable online contents accessible from the web site.

21. The at least one computer-readable medium of claim 17 , wherein the ascertaining comprises ascertaining the predicted popularity value based on a growth profile model of historical popularity values of other user-selectable online contents accessible from the web site.