Computing token-dependent affective response baseline levels utilizing a database storing affective responses

• US 8,938,403 B2
• Filed: 06/25/2011
• Issued: 01/20/2015
• Est. Priority Date: 05/11/2011
• Status: Active Grant

• Current Assignee: Affectomatics ltd.
• Original Assignee: Anat Thieberger Ben-Haim, Ari M. Frank, Gil Thieberger
• Inventors: Frank, Ari M., Thieberger, Gil, Thieberger Ben-Haim, Anat
• Primary Examiner(s): Chang, Li-Wu
• Assistant Examiner(s): FINK, THOMAS ANDREW
• Application Number:
  US13/168,967
• Publication Number:
  US 20120290515A1
• Time in Patent Office: 1,305 Days
• Field of Search: 706/12
• US Class Current: 706/12
• CPC Class Codes: G06F 3/013 Eye tracking input arrangements G06F3/015 takes precedence
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

Computing token-dependent affective response baseline levels for a user involves: receiving a certain temporal window of token instances, receiving from a database affective response annotations corresponding to temporal windows of token instances for which a difference from the certain temporal window of token instances is below a threshold, and computing a baseline level based on the affective response annotations. Wherein the baseline level includes a value that represents at least one of: an emotional response, and a value of a user measurement channel.

Claims

1. A method for computing token-dependent affective response baseline levels for a user, comprising:

receiving a certain temporal window of token instances;

receiving, from a database, first and third affective response annotations corresponding to first and third temporal windows of token instances, respectively;

wherein the database further stores a second affective response annotation corresponding to a second temporal window of token instances, the user was exposed to the first, second, and third, temporal windows of token instances during first, second, and third time periods, respectively; and

wherein the first time period precedes the second time period, and the second time period precedes the third time period;

using a distance function, calculating a first metric, a second metric, and a third metric between the certain temporal window of token instances and of the first temporal window of token instances, the second temporal window of token instances, and the third temporal window of token instances, respectively;

wherein the first and third metrics are below a predefined threshold, while the second metric is not below the predefined threshold;

computing a first affective response baseline level associated with the certain temporal window of token instances based on data comprising the first and third affective response annotations;

receiving an additional temporal window of token instances;

receiving, from the database, the second affective response annotation;

wherein a fourth metric between the additional temporal window of token instances and the second temporal window of token instances is below the predefined threshold; and

computing a second affective response baseline level associated with the additional temporal window of token instances based on data comprising the second affective response annotation;

wherein the computed first affective response baseline level is different from the computed second affective response baseline level;

wherein the first and second affective response baseline levels each represent a value comprising at least one of;

an emotional response, and a value of a user measurement channel.

2. The method of claim 1, further comprising utilizing the second affective response annotation for computing the first affective response baseline level;

wherein, for purpose of computing the first affective response baseline level, the second affective response annotation is assigned lower weight than weight assigned to at least one of the first and third affective response annotations.

3. The method of claim 1, further comprising generating a sample from data comprising:

the affective response first baseline level and the certain temporal window of token instances; and

providing the sample to a machine learning predictor to predict affective response of the user to the certain temporal window of token instances.

4. The method of claim 1, wherein the first, second, and third temporal windows of token instances comprise first, second, and third situation identifiers, respectively;

and wherein the first and third situation identifiers correspond to a first situation, and the second situation identifier corresponds to a second situation that is different than the first situation.

5. The method of claim 1, wherein the first and third temporal windows of token instances comprise token instances generated by a first token source, and the second temporal window of token instances comprises token instances generated by a second token source, which is different from the first token source.

6. The method of claim 1, wherein at least one of the first, second, and third temporal windows of token instances comprises at least two different token instances with overlapping instantiation periods.

7. The method of claim 1, wherein computing the first affective response baseline level comprises performing at least one of the following to values comprising the first and third affective response annotations:

averaging the values, applying low-pass filtering to the values, applying Fourier transform to the values, and performing wavelet transform analysis on the values.

8. A system configured to compute token-dependent affective response baseline levels for a user, comprising:

a database comprising first, second, and third affective response annotations corresponding to first, second, and third temporal windows of token instances, respectively;

wherein, the user was exposed to the first, second, and third, temporal windows of token instances during first, second, and third time periods, respectively; and

wherein the first time period precedes the second time period, and the second time period precedes the third time period; and

a baseline calculator, implemented utilizing a processor, configured to receive a certain temporal window of token instances and to compute, using a distance function, a first metric, a second metric, and a third metric between the certain temporal window of token instances and of the first temporal window of token instances, the second temporal window of token instances, and the third temporal window of token instances respectively;

wherein the first and third metrics are below a predefined threshold, while the second metric is not below the predefined threshold;

the baseline calculator is further configured to receive, from the database, the first and third affective response annotations and to compute a first affective response baseline level, associated with the certain temporal window of token instances, based on data comprising the first and third affective response annotations;

the baseline calculator is further configured to receive an additional temporal window of token instances;

the baseline calculator is further configured to receive, from the database, the second affective response annotation;

wherein a fourth metric between the additional temporal windows of token instances and the second temporal windows of token instances is below the predefined threshold;

the baseline calculator is further configured to compute a second affective response baseline level associated with the additional temporal window of token instances based on data comprising the second affective response annotation;

wherein the computed first affective response baseline level is different from the computed second affective response baseline level;

wherein the first and second affective response baseline levels each represent a value comprising at least one of;

an emotional response, and a value of a user measurement channel.

9. The system of claim 8, wherein the baseline calculator is further configured to utilize the second affective response annotation to compute the first affective response baseline level;

wherein, for purpose of computing the first affective response baseline level, the second affective response annotation is assigned lower weight than weight assigned to the first and third affective response annotations.

10. The system of claim 8, further comprising a sample generator configured to generate a sample from data comprising:

the first affective response baseline level, and the certain temporal window of token instances;

the sample generator is further configured to provide the sample to a machine learning predictor to predict affective response of the user to the certain temporal window of token instances.

11. The system of claim 8, wherein the first, second, and third temporal windows of token instances comprise first, second, and third situation identifiers, respectively;

and wherein the first and third situation identifiers correspond to a first situation, and the second situation identifier corresponds to a second situation that is different than the first situation.

12. The system of claim 8, wherein the first and third temporal windows of token instances comprise token instances generated by a first token source, and the second temporal window of token instances comprises token instances generated by a second token source, which is different from the first token source.

13. The system of claim 8, wherein at least one of the first, second, and third temporal windows of token instances comprises at least two different token instances with overlapping instantiation periods.

14. The system of claim 8, wherein for computing the first affective response baseline level the baseline calculator is further configured to perform at least one of the following to values comprising the first and third affective response annotations:

averaging the values, applying low-pass filtering to the values, applying Fourier transform to the values, and performing wavelet transform analysis on the values.

15. A non-transitory computer readable medium for use in a computer to compute token-dependent affective response baseline levels for a user;

the computer includes a processor, and the non-transitory computer readable medium comprising;

program code for receiving a certain temporal window of token instances;

program code for receiving, from a database, first and third affective response annotations corresponding to first and third temporal windows of token instances, respectively;

wherein the database further stores a second affective response annotation corresponding to a second temporal window of token instances, the user was exposed to the first, second, and third, temporal windows of token instances during first, second, and third time periods, respectively; and

wherein the first time period precedes the second time period, and the second time period precedes the third time period;

program code for calculating, using a distance function, a first metric, a second metric, and a third metric between the certain temporal window of token instances and the first temporal window of token instances, the second temporal window of token instances, and the third temporal window of token instances, respectively;

wherein the first and third metrics are below a predefined threshold, while the second metric is not below the predefined threshold;

program code for computing a first affective response baseline level associated with the certain temporal window of token instances based on data comprising the first and third affective response annotations;

program code for receiving an additional temporal window of token instances;

program code for receiving, from the database, the second affective response annotation;

wherein a fourth metric between the additional temporal window of token instances and the second temporal window of token instances is below the predefined threshold; and

program code for computing a second affective response baseline level associated with the additional temporal window of token instances based on data comprising the second affective response annotation;

wherein the computed first affective response baseline level is different from the computed second affective response baseline level;

wherein the first and second affective response baseline levels each represent a value comprising at least one of;

an emotional response, and a value of a user measurement channel.

16. The non-transitory computer readable medium of claim 15, further comprising program code for utilizing the second affective response annotation for computing the first affective response baseline level;

wherein, for purpose of computing the first baseline level, the second affective response annotation is assigned lower weight than weight assigned to at least one of the first and third affective response annotations.

17. The non-transitory computer readable medium of claim 15, further comprising program code for generating a sample from data comprising:

the first affective response baseline level and the certain temporal window of token instances; and

providing the sample to a machine learning predictor to predict affective response of the user to the certain temporal window of token instances.

18. The non-transitory computer readable medium of claim 15 wherein the first, second, and third temporal windows of token instances comprise first, second, and third situation identifiers, respectively;

and wherein the first and third situation identifiers correspond to a first situation, and the second situation identifier corresponds to a second situation that is different than the first situation.

19. The non-transitory computer readable medium of claim 15, wherein the first and third temporal windows of token instances comprise token instances generated by a first token source, and the second temporal window of token instances comprises token instances generated by a second token source, which is different from the first token source.

20. The non-transitory computer readable medium of claim 15, wherein computing the first affective response baseline level comprises performing at least one of the following to values comprising the first and third affective response annotations:

averaging the values, applying low-pass filtering to the values, applying Fourier transform to the values, and performing wavelet transform analysis on the values.