Content pre-fetching for computing devices

• US 8,849,731 B2
• Filed: 02/23/2012
• Issued: 09/30/2014
• Est. Priority Date: 02/23/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Lymberopoulos, Dimitrios, Riva, Oriana, Strauss, Karin
• Primary Examiner(s): Holmes, Michael B
• Application Number:
  US13/403,827
• Publication Number:
  US 20130226837A1
• Time in Patent Office: 950 Days
• US Class Current: 706/12
• CPC Class Codes: G06F 16/9574 of access to content, e.g. by caching

Abstract

The subject disclosure is directed towards a technology that timely pre-fetches content to a computing device based upon a prediction that a user will be requesting access to the content. Features comprising temporal features, spatial features, spatiotemporal features and/or other features associated with content are provided to a model trained at least in part with historical access data. The model returns information from which a determination of whether to pre-fetch the content is made.

Claims

1. In a computing environment, a method performed at least in part on at least one processor, comprising:

determining whether to pre-fetch content to a computing device, including generating features comprising temporal-related feature data associated with the content, providing the features as a feature vector input into a model trained at least in part with historical access data, and obtaining information from the model indicative of whether to pre-fetch the content.

2. The method of claim 1 wherein the historical access data corresponds to a single user, and wherein a piece of content represented by the feature vector corresponds to a URL from a set of one or more URLs that are identified as targeted URLs based upon the historical access data of the user.

3. The method of claim 1 further comprising computing the feature vector including data representative of spatial-related feature data.

4. The method of claim 1 further comprising computing the feature vector including data representative of spatiotemporal-related feature data.

5. The method of claim 1 further comprising computing the feature vector including data representative of popularity-related feature data.

6. The method of claim 1 further comprising computing the feature vector including data representative of context-related feature data.

7. The method of claim 1 further comprising computing the feature vector including data representative of the temporal-related feature data, spatial-related feature data and spatiotemporal-related feature data.

8. The method of claim 1 wherein generating the features comprises processing a set of one or more URLs that are identified as targeted URLs based upon the historical access data.

9. The method of claim 1 wherein obtaining the information from the model indicative of whether to pre-fetch the content comprises receiving a predicted probability value, and further comprising, pre-fetching the content when the predicted probability value achieves a threshold probability value.

10. A system comprising, a model trained based upon historical access data to predict whether to pre-fetch content, the model configured to receive as input a feature vector that represents a piece of content, the feature vector including data representative of spatiotemporal-related feature data associated with the piece of content, the model further configured to process the feature vector to output information indicative of whether a user is likely to request timely access to the content.

11. The system of claim 10 wherein the model is run on a computing device to which content is pre-fetched.

12. The system of claim 10 wherein at least part of the historical access data model is maintained on a computing device to which content is pre-fetched.

13. The system of claim 10 wherein the historical access data corresponds to a single user, and wherein the piece of content corresponds to a URL from a set of one or more URLs that are identified as targeted URLs based upon the historical access data of the user.

14. The system of claim 10 wherein the feature vector further includes data representative of spatial-related feature data or temporal-related feature data, or both spatial-related feature data and temporal-related feature data.

15. The system of claim 10 wherein the feature vector further includes data representative of popularity-related feature data or context-related feature data, or both popularity-related feature data and context-related feature data.

16. The system of claim 10 wherein the model is further trained based upon feedback indicative of whether the user actually requested timely access to the content.

17. One or more computer-readable media having computer-executable instructions, which when executed perform steps, comprising:

selecting a content identifier;

generating a feature vector for the content identifier, the feature vector including temporal-related featured data, spatial-related feature data and spatiotemporal-related feature data associated with the content identifier;

providing the feature vector to a model;

receiving a prediction-related value from the model representative of the feature vector; and

pre-fetching content corresponding to the content identifier to a computing device if the prediction-related value indicates that future access to the content is likely to be requested.

18. The one or more computer-readable media of claim 17 wherein pre-fetching the content comprises batching a request for the content with at least one other request for other content.

19. The one or more computer-readable media of claim 17 having further computer-executable instructions comprising, training the model based upon historical access data.

20. The one or more computer-readable media of claim 17 having further computer-executable instructions comprising, monitoring accuracy with respect to predicted results and actual results, and adjusting the pre-fetching based upon the monitoring.