CONTEXT LABELS FOR DATA CLUSTERS

• US 20140129560A1
• Filed: 03/14/2013
• Published: 05/08/2014
• Est. Priority Date: 11/02/2012
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: GROKOP, Leonard Henry, SADASIVAM, Shankar
• Granted Patent: US 9,740,773 B2
• Time in Patent Office: Days
• US Class Current: 707/737
• CPC Class Codes: G05B 13/00 Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion G05B19/00 takes precedence
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/337 Profile generation, learning or modification
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/58 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/043 Distributed expert systems; Blackboards
• CPC Class Codes: H04M 1/72454 according to context-related or environment-related conditions

Abstract

Systems and methods for applying and using context labels for data clusters are provided herein. A method described herein for managing a context model associated with a mobile device includes obtaining first data points associated with a first data stream assigned to one or more first data sources; assigning ones of the first data points to respective clusters of a set of clusters such that each cluster is respectively assigned ones of the first data points that exhibit a threshold amount of similarity and are associated with times within a threshold amount of time of each other; compiling statistical features and inferences corresponding to the first data stream or one or more other data streams assigned to respective other data sources; assigning context labels to each of the set of clusters based on the statistical features and inferences.

Claims

1. A method for managing a context model associated with a mobile device, the method comprising:

obtaining first data points associated with a first data stream assigned to one or more first data sources;

assigning ones of the first data points to respective clusters of a set of clusters such that each cluster is respectively assigned ones of the first data points that exhibit a threshold amount of similarity and are associated with times within a threshold amount of each other;

compiling statistical features and inferences corresponding to the first data stream or one or more other data streams assigned to respective other data sources; and

assigning context labels to each of the set of clusters based on the statistical features and inferences.

2. The method of claim 1 wherein the first data sources or the other data sources comprise at least one of an audio sensor, a location sensor, a network sensor, a motion sensor, a camera, or a calendar.

3. The method of claim 1 wherein the statistical features comprise at least one of satellite positioning fixes, time measurements, weather data, light sensor readings, proximity sensor readings, camera data, proximate points of interest, Bluetooth measurement, Wi-Fi measurements, ambient audio level, or movement detection.

4. The method of claim 1 wherein the statistical inferences comprise at least one of motion state, device position, speech detection, speaker recognition, target sound detection, or location detection.

5. The method of claim 1 wherein the first data sources comprise a plurality of sensors.

6. The method of claim 1 further comprising generating a context model using the context labels and the set of clusters.

7. The method of claim 6 further comprising:

obtaining second data points associated with a second data stream assigned to one or more second data sources;

determining a cluster of the set of clusters that corresponds to the second data points; and

selecting a context label associated with the determined cluster.

8. The method of claim 1 wherein assigning the context labels comprises selecting context labels from a set of context labels for each of the set of clusters.

9. The method of claim 8 further comprising:

identifying at least one cluster among the set of clusters having less than a threshold degree of relation to any context label associated with the set of clusters; and

creating a new context label for the at least one cluster.

10. The method of claim 8 further comprising:

identifying at least one cluster among the set of clusters having less than a threshold degree of relation to any context label associated with the set of clusters; and

assigning an unknown context label to the at least one cluster.

11. The method of claim 1 wherein:

the obtaining comprises identifying a user of the mobile device;

the first data points comprise data corresponding to the user of the mobile device; and

the method further comprises creating a user context model for the user of the mobile device based on the context labels and the set of clusters.

12. The method of claim 1 wherein:

the first data points comprise data corresponding to a first user of the mobile device and at least one second user of the mobile device or one or more other devices; and

the method further comprises creating a multi-user context model for the first user and the at least one second user based on the context labels and the set of clusters.

13. A method for performing a context inference based on a context model, the method comprising:

retrieving the context model, the context model comprising sensor data points temporally grouped into respective ones of a set of clusters and context labels assigned to each of the set of clusters;

obtaining first data points associated with a first data stream assigned to one or more first data sources;

determining at least one cluster of the context model that is representative of the first data points; and

selecting an output context label associated with the at least one determined cluster.

14. The method of claim 13 wherein the determining comprises assigning confidence levels to a plurality of clusters among the set of clusters that correspond to the first data points.

15. The method of claim 14 wherein the selecting comprises selecting a context label corresponding to a cluster of the context model having a highest confidence level.

16. The method of claim 14 wherein the selecting comprises generating a vector of respective ones of the set of clusters and their respective confidence levels.

17. The method of claim 14 further comprising collecting additional data from at least one sensor when one or more of the confidence levels are below a threshold.

18. The method of claim 13 wherein the selecting comprises:

directing collection of additional data from the one or more first data sources; and

selecting the output context label based on the additional data collected from the one or more first data sources in response to the directing.

19. The method of claim 13 further comprising obtaining second data points associated with a second data stream assigned to one or more second data sources, wherein the determining comprises determining clusters of the context model that correspond to the first data points and the second data points.

20. The method of claim 19 wherein the selecting comprises:

selecting a first context label associated with a first cluster that is representative of the first data points;

selecting a second context label associated with a second cluster that is representative of the second data points; and

determining the output context label by combining the first context label and the second context label.

21. The method of claim 19 further comprising:

performing a first context inference based on the selected output context label;

performing a second context inference based on instantaneous features and inferences derived from the second data sources; and

combining the first context inference and the second context inference to obtain a combined context inference.

22. An apparatus that facilitates managing an associated context model, the apparatus comprising:

one or more first data sources configured to provide first data points associated with a first data stream;

a clustering module communicatively coupled to the first data sources and configured to assign ones of the first data points to respective clusters of a set of clusters such that each cluster is respectively assigned ones of the first data points that exhibit a threshold amount of similarity and are associated with times within a threshold amount of each other;

a statistics module communicatively coupled to the first data sources and the clustering module and configured to compile statistical features and inferences for the first data stream or one or more other data streams assigned to respective other data sources; and

a context modeling module communicatively coupled to the clustering module and the statistics module and configured to assign context labels to one or more of the set of clusters based on the statistical features and inferences.

23. The apparatus of claim 22 wherein the first data sources comprise at least one of an audio sensor, a location sensor, a network sensor, a motion sensor, a camera, or a calendar.

24. The apparatus of claim 22 wherein the statistics module is further configured to determine statistical inferences comprising at least one of motion state, device position, speech detection, speaker recognition, target sound detection, or location detection.

25. The apparatus of claim 22 wherein the context modeling module is further configured to generate a context model using the context labels and the set of clusters.

26. The apparatus of claim 25 wherein:

the apparatus further comprises one or more second data sources communicatively coupled to the clustering module and configured to provide second data points associated with a second data stream,

the clustering module is further configured to determine a cluster of the context model that corresponds to the second data points, and

the apparatus further comprises a context inference module communicatively coupled to the context modeling module and the clustering module and configured to select a context label associated with the determined cluster.

27. The apparatus of claim 22 wherein the context modeling module is further configured to identify at least one cluster among the set of clusters having less than a threshold degree of relation to any context label associated with the set of clusters and to create a new context label for the at least one cluster.

28. The apparatus of claim 22 wherein the context modeling module is further configured to identify at least one cluster among the set of clusters having less than a threshold degree of relation to any context label associated with the set of clusters and to assign an unknown context label to the at least one cluster.

29. An apparatus for performing a context inference based on a context model, the apparatus comprising:

a context modeling module configured to provide a context model, the context model comprising sensor data points temporally grouped into respective ones of a set of clusters and context labels assigned to each of the set of clusters;

one or more first data sources configured to provide first data points associated with a first data stream; and

a context inference module communicatively coupled to the context modeling module and the one or more first data sources and configured to determine at least one cluster of the context model that is representative of the first data points and to select an output context label associated with the at least one determined cluster.

30. The apparatus of claim 29 wherein the context inference module is further configured to assign confidence levels to a plurality of clusters among the set of clusters that correspond to the first data points.

31. The apparatus of claim 30 wherein the context inference module is further configured to select a context label corresponding to a cluster of the context model having a highest confidence level.

32. The apparatus of claim 29 further comprising one or more second data sources communicatively coupled to the context inference module and configured to provide second data points associated with a second data stream, wherein the context inference module is further configured to determine clusters of the context model that correspond to the first data points and the second data points.

33. The apparatus of claim 32 wherein the context inference module is further configured to select a first context label associated with a first cluster that is representative of the first data points, to select a second context label associated with a second cluster that is representative of the second data points, and to determine the output context label by combining the first context label and the second context label.

34. The apparatus of claim 32 wherein the context inference module is further configured to perform a first context inference based on the selected output context label, to perform a second context inference based on instantaneous features and inferences derived from the second data sources, and to combine the first context inference and the second context inference to obtain a combined context inference.

35. The apparatus of claim 29 wherein the first data sources comprise at least one of an audio sensor, a location sensor, a network sensor, a motion sensor, a camera, or a calendar.

36. An apparatus for managing a context model associated with a mobile device, the apparatus comprising:

means for obtaining first data points associated with a first data stream assigned to one or more first data sources;

means for assigning ones of the first data points to respective clusters of a set of clusters such that each cluster is respectively assigned ones of the first data points that exhibit a threshold amount of similarity and are associated with times within a threshold amount of each other;

means for compiling statistical features and inferences for the first data stream or one or more other data streams assigned to respective other data sources; and

means for associating one or more of the set of clusters with context labels based on the statistical features and inferences.

37. The apparatus of claim 36 further comprising means for generating a context model using the context labels and the set of clusters.

38. The apparatus of claim 37 wherein:

the apparatus further comprises means for obtaining second data points associated with a second data stream assigned to one or more second data sources;

the means for assigning comprises means for determining a cluster of the context model that corresponds to the second data points; and

the apparatus further comprises means for selecting at least one of the context labels that is associated with the determined cluster.

39. The apparatus of claim 36 further comprising:

means for identifying at least one cluster among the set of clusters having less than a threshold degree of relation to any context label associated with the set of clusters; and

means for creating a new context label for the at least one cluster or assigning an unknown context label to the at least one cluster.

40. An apparatus for performing a context inference based on a context model, the apparatus comprising:

means for retrieving a context model, the context model comprising sensor data points temporally grouped into respective ones of a set of clusters and context labels assigned to each of the set of clusters;

means for obtaining first data points from one or more first data sources associated with a first data stream;

means for determining at least one cluster of the context model that is representative of the first data points; and

means for selecting an output context label associated with the at least one determined cluster.

41. The apparatus of claim 40 further comprising means for obtaining second data points from one or more second data sources associated with a second data stream, wherein the means for determining comprises means for determining clusters of the context model that correspond to the first data points and the second data points.

42. The apparatus of claim 41 wherein the means for selecting comprises:

means for selecting a first context label associated with a first cluster that is representative of the first data points;

means for selecting a second context label associated with a second cluster that is representative of the second data points; and

means for selecting the output context label by combining the first context label and the second context label.

43. The apparatus of claim 41 further comprising:

means for performing a first context inference based on the output context label;

means for performing a second context inference based on instantaneous features and inferences derived from the second data sources; and

means for combining the first context inference and the second context inference to obtain a combined context inference.

44. The apparatus of claim 40 wherein the first data sources comprise at least one of an audio sensor, a location sensor, a network sensor, a motion sensor, a camera, or a calendar.

45. A processor-executable computer storage medium comprising processor-executable instructions configured to cause a processor to:

obtain first data points associated with a first data stream assigned to one or more first data sources;

assign ones of the first data points to respective clusters of a set of clusters such that each cluster is respectively assigned ones of the first data points that exhibit a threshold amount of similarity and are associated with times within a threshold amount of each other;

compile statistical features and inferences for the first data stream or one or more other data streams assigned to respective other data sources; and

associate one or more of the set of clusters with context labels based on the statistical features and inferences.

46. The computer storage medium of claim 45 further comprising instructions for causing the processor to generate a context model using the context labels and the set of clusters.

47. The computer storage medium of claim 46 further comprising instructions configured to cause the processor to:

obtain second data points associated with a second data stream assigned to one or more second data sources;

determine a cluster of the context model that corresponds to the second data points; and

select at least one of the context labels that is associated with the determined cluster.

48. A processor-executable computer storage medium comprising processor-executable instructions configured to cause a processor to:

retrieve a context model, the context model comprising sensor data points temporally grouped into respective ones of a set of clusters and context labels assigned to each of the set of clusters;

obtain first data points from one or more first data sources associated with a first data stream;

determine at least one cluster of the context model that is representative of the first data points; and

select an output context label associated with the at least one determined cluster.

49. The computer storage medium of claim 48 further comprising instructions configured to cause the processor to:

obtain second data points from one or more second data sources associated with a second data stream; and

determine clusters of the context model that correspond to the first data points and the second data points.

50. The computer storage medium of claim 49 further comprising instructions configured to cause the processor to:

select a first context label associated with a first cluster that is representative of the first data points;

select a second context label associated with a second cluster that is representative of the second data points; and

select the output context label by combining the first context label and the second context label.

51. The computer storage medium of claim 49 further comprising instructions configured to cause the processor to:

perform a first context inference based on the output context label;

perform a second context inference based on instantaneous features and inferences derived from the second data sources; and

combine the first context inference and the second context inference to obtain a combined context inference.