Method and system for detecting non-powered video playback devices

• US 8,863,166 B2
• Filed: 04/06/2011
• Issued: 10/14/2014
• Est. Priority Date: 04/06/2011
• Status: Active Grant

• Current Assignee: comScore, Inc.
• Original Assignee: comScore, Inc.
• Inventors: Harsh, Aaron, Yazdani, Amir, Vinson, Michael, Goerlich, Bruce
• Primary Examiner(s): CHEN, CAI Y
• Application Number:
  US13/081,437
• Publication Number:
  US 20120260280A1
• Time in Patent Office: 1,287 Days
• Field of Search: None
• US Class Current: 725/21
• CPC Class Codes: H04N 21/24 Monitoring of processes or resources, e.g. monitoring of server load, available bandwidth, upstream requests monitoring of server performance or load G06F11/34; arrangements for observation, testing or troubleshooting for broadcast or for distribution combined with broadcast H04H20/12
• CPC Class Codes: H04N 21/252 Processing of multiple end-users' preferences to derive collaborative data
• CPC Class Codes: H04N 21/44224 Monitoring of user activity on external systems, e.g. Internet browsing
• CPC Class Codes: H04N 21/658 Transmission by the client directed to the server
• CPC Class Codes: H04N 21/81 Monomedia components thereof
• CPC Class Codes: H04N 5/91 Television signal processing therefor

Abstract

A method and system for identifying tune data from set top boxes associated with televisions, video monitors, or other video playback devices that are likely powered off. In some embodiments, survival curves are constructed that predict a length of time before a video playback device is powered off after a tuning event. The survival curves are used to predict the likelihood that a video playback device is powered off. Viewership estimates made from tune data reported from set top boxes can be adjusted to take account of the video playback devices that are predicted to be powered off.

Claims

1. A computing system for analyzing video playback device viewing data, comprising:

a memory for storing programmed instructions;

a processor configured to execute the programmed instructions to perform operations including;

maintaining multiple survival models, each survival model associated with a portion of a day, of a plurality of portions of the day, and comprising a plurality of probabilities in the open interval between 0 and 1 that a video playback device associated with a set top box is powered off as a function of time following a tuning event, each survival model generated from tune data from a plurality of set top boxes, the tune data associated with the corresponding portion of the day;

receiving tune data indicative of content tuned to by a plurality of set top boxes, the tune data including a plurality of tuning events, each tuning event including a time of the tuning event and a corresponding content identifier that was tuned at the tuning event; and

analyzing the tune data from the plurality of set top boxes to estimate video viewership, the analysis including, for each set top box of the plurality of set top boxes;

selecting a survival model from the multiple survival models, the selected survival model corresponding to a same portion of the day as reflected by at least one of the tuning events from the set top box;

selecting a probability that the video playback device associated with the set top box was powered off;

determining a time corresponding to the probability using the selected survival model; and

excluding, from the analysis to estimate video viewership, tuning data from the set top box that occurs after the determined time.

2. The computing system of claim 1, wherein the survival models are generated by:

analyzing tune data from the plurality of set top boxes to assess when associated video playback devices are powered off; and

compiling statistics predicting the length of time before a video playback device is powered off after a tuning event, wherein the statistics are compiled from the analyzed tune data.

3. The computing system of claim 2, wherein the statistics regarding a length of time before a video playback device is powered off are compiled by detecting power off signals in the tune data from the plurality of set top boxes.

4. The computing system of claim 2, wherein the statistics regarding a length of time before a video playback device is powered off are compiled by:

identifying a number of reliable set top boxes in the tune data; and

detecting power off signals in the tune data from the reliable set top boxes.

5. The computing system of claim 2, wherein the statistics that predict a length of time before a video playback device is powered off are compiled by analyzing the tune data for times between two successive tuning events and a power off signal received from a video playback device between those two successive tuning events.

6. The computing system of claim 5, wherein the statistics are compiled for different times of a day.

7. The computing system of claim 5, wherein the statistics are compiled for different days of a week.

8. The computing system of claim 5, wherein the statistics are compiled for different months of a year.

9. The computing system of claim 5, wherein the statistics are compiled for types of broadcast programs.

10. The computing system of claim 5, wherein the statistics are compiled for types of networks.

22. The computing system of claim 1, wherein selecting a survival model from the multiple survival models, the selected survival model corresponding to a same portion of the day as reflected by at least one of the tuning events from the set top box comprises determining that a beginning time of the at least one survival model is closer to the first tuning event of the tuning events than beginning times of the other multiple survival models.

25. The computing system of claim 1, wherein selecting a probability comprises a portion of a Monte Carlo technique to guarantee that a probability distribution function of adjusted tune lengths matches an empirical distribution determined from reliable set top boxes.

11. A non-transitory, computer readable media with instructions stored thereon that are executable by a processor to perform operations of:

maintaining multiple survival models, each survival model associated with a portion of a day, of a plurality of portions of the day, and comprising a plurality of probabilities in the open interval between 0 and 1 that a video playback device associated with a set top box is powered off as a function of time following a tuning event, each survival model generated from tune data from a plurality of set top boxes, the tune data associated with the corresponding portion of the day;

receiving tune data indicative of content tuned to by a plurality of set top boxes, the tune data including an identification of each of the set top boxes of the plurality of set top boxes and a plurality of tuning events, each tuning event including a time of the tuning event and a corresponding content identifier that was tuned at the tuning event; and

analyzing the tune data from the plurality of set top boxes to estimate video viewership, the analysis including, for each set top box of the plurality of set top boxes;

selecting a survival model from the multiple survival models, the selected survival model corresponding to a same portion of the day as reflected by at least one of the tuning events from the set top box;

selecting a probability that the video playback device associated with the set top box was powered off;

determining a time corresponding to the probability using the at least one survival model; and

excluding, from the analysis to estimate video viewership, tuning data from the set top box that occurs after the determined time.

12. The non-transitory computer readable media of claim 11, wherein the survival models are generated by:

analyzing tune data from the plurality of set top boxes to assess when associated video playback devices are powered off; and

compiling statistics predicting the length of time before a video playback device is powered off after a tuning event, wherein the statistics are compiled from the analyzed tune data.

13. The non-transitory computer readable media of claim 12, wherein compiling statistics predicting a length of time before a video playback device is powered off after a tuning event includes the operation of detecting power off signals in the tune data.

14. The non-transitory computer readable media of claim 12, wherein compiling statistics predicting a length of time before a video playback device is powered off after a tuning event includes the operations of:

identifying a number of reliable set top boxes in the tune data; and

detecting power off signals in the tune data from the reliable set top boxes.

23. The non-transitory, computer readable media of claim 11, wherein selecting a survival model from the multiple survival models, the selected survival model corresponding to a same portion of the day as reflected by at least one of the tuning events from the set top box comprises determining that a beginning time of the at least one survival model is closer to the first tuning event of the tuning events than beginning times of the other multiple survival models.

26. The non-transitory, computer readable media of claim 11, wherein selecting a probability comprises a portion of a Monte Carlo technique to guarantee that a probability distribution function of adjusted tune lengths matches an empirical distribution determined from reliable set top boxes.

15. A computer-implemented method of analyzing tune data, comprising:

maintaining multiple survival models, each survival model associated with a portion of a day, of a plurality of portions of the day, and comprising a plurality of probabilities in the open interval between 0 and 1 that a video playback device associated with a set top box is powered off as a function of time following a tuning event, each survival model generated from tune data from a plurality of set top boxes, the tune data associated with the corresponding portion of the day;

receiving with the computer, tune data from a plurality of set top boxes, the tune data including a plurality of tuning events, each tuning event including a time of the tuning event and a corresponding content identifier that was tuned at the tuning event; and

analyzing, with the computer, the tune data from the plurality of set top boxes to estimate video viewership, the analysis including, for each set top box of the plurality of set top boxes;

selecting a survival model from the multiple survival models, the selected survival model corresponding to a same portion of the day as reflected by at least one of the tuning events from the set top box;

selecting a probability that the video playback device associated with the set top box was powered off;

determining a time corresponding to the probability using the at least one survival model; and

excluding from at least a portion of an analysis the tuning data from the set top box that occurs after the determined time.

16. The computer-implemented method of claim 15, wherein the multiple survival models are generated by:

constructing, with the computer, survival curves that indicate a likelihood that a video playback device is powered off within a time interval after a tuning event, wherein the survival curves are constructed from set top boxes that report a power off event in the tune data; and

using the survival curves to identify, with the computer, one or more set top boxes that do not report power off events in the tune data and that are associated with video playback devices that are likely powered off.

17. The computer-implemented method of claim 16, wherein the survival curves are constructed by the computer for different hours of the day.

18. The computer-implemented method of claim 16, wherein the survival curves are constructed by a computer for different days of the week or months of the year.

19. The computer-implemented method of claim 16, wherein the survival curves are constructed by a computer for different types of broadcast programs.

20. The computer-implemented method of claim 16, wherein the survival curves are constructed by a computer for different types of networks.

21. The computer-implemented method of claim 16, wherein a computer adjusts viewership estimates for the tune data from set top boxes associated with video playback devices that are identified as likely being powered off.

24. The computer-implemented method of claim 15, wherein selecting a survival model from the multiple survival models, the selected survival model corresponding to a same portion of the day as reflected by at least one of the tuning events from the set top box comprises determining that a beginning time of the at least one survival model is closer to the first tuning event of the tuning events than beginning times of the other multiple survival models.

27. The computer-implemented method of claim 15, wherein selecting a probability comprises a portion of a Monte Carlo technique to guarantee that a probability distribution function of adjusted tune lengths matches an empirical distribution determined from reliable set top boxes.