Network based content transmission based on client device parameters

1. A system for network based content transmission based on client device parameters, comprising:

a data processing system having one or more processors to:

receive, from a client device, a request for content for insertion into a first content source, the request including: (1) a device identifier associated with the client device, and (2) a keyword of the first content source received by the client device at a first time;

identify, from a device profile database, a keyword of a second content source received by the client device at a second time prior to the first time;

select, for analysis, a content item based on the keyword of the first content source, the keyword of the second content source, and a category identifier associated with the content item;

estimate a predicted likelihood of interaction for the content item using a prediction model and the keyword of the second content source, an interaction frequency metric indicating a frequency of interaction with content of the second content source, and an amount of time the keyword of the second content source has been stored in the device profile database in association with the device identifier;

select, for provision to the client device, the content item based on the predicted likelihood of interaction; and

provide, to the client device, the content item for presentation on the content source.

2. The system of claim 1 , wherein the data processing system estimates the predicted likelihood of interaction for the content item based on the prediction model, the prediction model using the keyword of the second content source, the interaction frequency metric, and the amount of time the keyword of the second content source has been stored in the device profile database in association with the device identifier as a set of coefficients in the prediction model.

3. The system of claim 1 , wherein the data processing system is further configured to:

perform an image recognition to the content source to determine the keyword of the first content source; and

perform at least one of the image recognition and a text recognition to the content item to determine the category identifier.

4. The system of claim 1 , wherein the data processing system is further configured to:

parse the first content source to identify a first keyword of the first content source;

match the first keyword to a keyword set, the keyword set associated with one category identifier; and

identify the keyword of the first content source based on the match of the keyword to the keyword set.

5. The system of claim 1 , wherein the data processing system is further configured to:

select a plurality of candidate content items based on a match between the keyword of the first content source, the keyword of the second content source, and the category identifier, the plurality of content items numbering a predetermined limit; and

select, from the plurality of candidate content items, the content item based on the predicted likelihood of interaction estimated based on the prediction model.

6. The system of claim 1 , wherein the data processing system is further configured to:

receive, from the device, an indication of an interaction by the client device with the content item on the first content source; and

update the prediction model based on the receipt of the indication of the interaction by the client device.

7. The system of claim 1 , wherein the data processing system is further configured to:

receive, from the device, an indication of an interaction by the client device with the content item on the first content source; and

update a set of coefficients for the prediction model, the set of coefficients corresponding to the input values of the keyword of the second content source, the interaction frequency metric, and the amount of time the keyword of the second content source has been stored in the device profile database in association with the device identifier.

8. The system of claim 1 , comprising wherein the data processing system is further configured to:

determine a frequency for a keyword on the first content source, the frequency indicating a number of occurrences of the keyword on the first content source; and

identify the frequency for the keyword as one of the input values for the prediction model to estimate the predicted likelihood of interaction for the content item.

9. The system of claim 1 , wherein the predicted likelihood of interaction includes at least one of a predicted conversion rate and a predicted click-through rate.

10. The system of claim 1 , wherein the predicted likelihood of interest for the content item is determined based on at least one of a logistic regression technique, a linear regression technique, an artificial neural network technique, and a Bayesian classifier technique.

11. A method of estimating likelihood of interaction with content items on webpages in a computer networked environment, comprising:

receiving, by a data processing system having one or more processors, from a client device, a request for content for insertion into a first content source, the request including: (1) a device identifier corresponding to the client device, and (2) a keyword of the first content source received by the client device at a first time;

identifying, by the data processing system, from a device profile database, a keyword of a second content source received by the client device at a second time prior to the first time;

selecting, by the data processing system for analysis, a content item based on the keyword of the first content source, the keyword of the second content source, and a category identifier associated with the content item;

estimating, by the data processing system, a predicted likelihood of interaction for the content item using the prediction model, the prediction model using the input values of the second category identifier, the interaction frequency metric, and the-amount of time the keyword of the content source previously received by the client device has been stored in a data structure in association with the device identifier;

selecting, by the data processing system for provisioning to the client device, the content item based on the estimated predicted likelihood of interaction; and

providing, by the data processing system, to the client device, the content item for presentation on the content source.

12. The method of claim 11 , further comprising:

estimating, by the data processing system, the predicted likelihood of interaction for the content item based on the prediction model, the keyword of the second content source, the interaction frequency metric, and the amount of time the keyword of the second content source has been stored in the device profile database in association with the device identifier as a set of coefficients in the prediction model.

13. The method of claim 11 , further comprising:

performing, by the data processing system, an image recognition to the first content source to determine the keyword of the first content source; and

performing, by the data processing system, at least one of the image recognition and a text recognition to the content item to determine the category identifier.

14. The method of claim 11 , further comprising:

parsing, by the data processing system, the first content source to identify a first keyword of the first content source;

matching, by the data processing system, the first keyword to a keyword set, the keyword set associated with one category identifier; and

identifying, by the data processing system, the keyword of the first content source based on the match of the keyword to the keyword set.

15. The method of claim 11 , further comprising:

selecting, by the data processing system, a plurality of candidate content items based on a match between the keyword of the first content source, the keyword of the second content source, and the category identifier, the plurality of content items numbering a predetermined limit; and

selecting, by the data processing system, from the plurality of candidate content items, the content item based on the predicted likelihood of interaction estimated based on the prediction model.

16. The method of claim 11 , further comprising:

receiving, by the data processing system, from the device, an indication of an interaction by the client device with the content item on the first content source; and

updating, by the data processing system, the prediction model based on the receipt of the indication of the interaction by the client device.

17. The method of claim 11 , further comprising:

receiving, by the data processing system, from the device, an indication of an interaction by the client device with the content item on the first content source; and

updating, by the data processing system, a set of coefficients for the prediction model, the set of coefficients corresponding to the input values of the keyword of the second content source, the interaction frequency metric, and the amount of time the keyword of the second content source has been stored in the device profile in association with the device identifier.

18. The method of claim 11 , further comprising:

determining, by the data processing system, a frequency for a keyword on the first content source, the frequency indicating a number of occurrence of the keyword on the first content source; and

identifying, by the data processing system, the frequency for the keyword as one of the input values for the prediction model for estimating the predicted likelihood of interaction for the content item.

19. The method of claim 11 , further comprising:

estimating, by the data processing system, the predicted likelihood of interaction for the content item, the predicted likelihood of interaction including at least one of a predicted conversion rate and a predicted click-through rate.

20. The method of claim 11 , further comprising:

estimating, by the data processing system, the predicted likelihood of interaction for the content item based on the prediction model, the prediction model including at least one of a logistic regression, a linear regression, an artificial neural network technique, and a Bayesian classifier.