Feature type spectrum technique
Sensors are used to generate sample set data representing objects in a sample set. A computer system analyzes the sample set data to determine the frequencies with which features in a feature set are observed in the objects in the sample set. An example of such output is a bar chart representing the frequency of observation of features in the feature set in a particular object. The feature output may be used to identify one or more obscure (i.e., low frequency) features in the particular object. Machine learning may be used to learn associations between sample set data and features in the feature set.
1. A method performed by at least one computer processor executing computer program instructions stored on a non-transitory computer-readable medium, the method comprising:
(A) generating, for each feature F in a plurality of features, a plurality of frequencies of observation of feature F in an object O 1 , wherein the plurality of features includes at least one physical feature and at least one use-based feature, comprising:
(A) (1) receiving first textual input from a first human;
(A) (2) receiving second textual input from a second human, wherein the first textual input differs from the second textual input;
(A) (3) mapping the first textual input and the second textual input to the same feature F 0 in the plurality of features; and
(A) (4) determining that the first textual input and the second textual input indicate that the object O 1 has feature F 0 ;
(B) generating output representing the plurality of frequencies of observation of each feature F in object O 1 ;
(C) identifying, based on the plurality of frequencies of observation of each feature F in object O 1 , a first subset of the plurality of features having frequencies satisfying a low frequency criterion, comprising:
(C) (1) generating, for each feature F in the plurality of features, a frequency count for feature F in object O 1 based on the plurality of frequencies of observation of feature F in object O 1 ; and
(C) (2) determining, for each feature F in the plurality of features, whether the frequency count for feature F satisfies the low frequency criterion, comprising:
determining that the feature count for a first one of the plurality of features satisfies the low frequency criterion; and
determining that the feature count for a second one of the plurality of features does not satisfy the low frequency criterion;
(D) automatically learning a first association between the first textual input and the feature F 0 , and storing first association data representing the first association;
(E) automatically learning a second association between the second textual input and the feature F 0 , and storing second association data representing the second association; and
wherein the frequency count for at least one feature F is equal to zero.
2. The method of claim 1 , wherein (A) comprises:
(A) (1) generating, for each feature F in the plurality of features, a first indication of whether object O 1 was observed to have feature F, thereby generating a first plurality of indications for object O 1 ;
(A) (2) generating, for each feature F in the plurality of features, a second indication of whether object O 1 was observed to have feature F; thereby generating a second plurality of indications for object O 1 ; and
(A) (3) generating the plurality of frequencies of observation of feature F in object O 1 based on the first and second pluralities of indications for object O 1 .
3. The method of claim 1 , wherein the output representing the plurality of frequencies of observation of feature F in object O 1 comprises a chart representing the plurality of frequencies of observation of feature F in object O 1 .
4. The method of claim 3 , wherein the chart comprises a bar chart.
5. The method of claim 3 , wherein the chart comprises a pie chart.
6. The method of claim 1 , wherein the low frequency criterion comprises a maximum value, wherein (C) (2) comprises determining, for each feature F in the plurality of features, whether the frequency count for feature F is less than the maximum value, and wherein the first subset comprises features in the plurality of features having frequencies less than the maximum value.
7. The method of claim 1 , further comprising:
(F) identifying, based on the plurality of frequencies of observation of feature F in object O 1 , a second subset of the plurality of features having frequencies satisfying a high frequency criterion.
8. The method of claim 7 , wherein the high frequency criterion comprises a minimum value, and wherein the second subset comprises features in the plurality of features having frequencies greater than the minimum value.
9. The method of claim 1 , further comprising:
(F) generating output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion.
10. The method of claim 9 , wherein the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion includes output representing the frequencies satisfying the low frequency criterion.
11. The method of claim 9 , wherein the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion comprises a chart.
12. The method of claim 11 , wherein the chart comprises a bar chart.
13. The method of claim 11 , wherein the chart comprises a pie chart.
14. The method of claim 9 , wherein the output representing the plurality of frequencies of observation of feature F in object O 1 includes the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion.
15. The method of claim 14 , wherein the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion comprises output emphasizing the first subset of the plurality of features.
16. The method of claim 1 , wherein the object O is a physical object.
17. The method of claim 1 :
wherein (D) comprises using a machine learning engine to automatically learn the first association; and
wherein (E) comprises using the machine learning engine to automatically learn the second association.
18. The method of claim 1 , further comprising:
(F) receiving third textual input from the first human;
(G) receiving fourth textual input from the first human; and
(H) determining, based on at least one of the first association and the second association, that the third textual input and the fourth textual input indicate that a second object O 2 has feature F 0 .
19. The method of claim 1 , wherein (H) comprises determining, based on the first association and the second association, that the third textual input and the fourth textual input indicate that a second object O 2 has feature F 0 .
20. The method of claim 1 , wherein the first association data and the second association data are the same data.
21. The method of claim 1 , wherein (A) further comprises:
(A) (5) receiving, from a first sensor, first sensor data representing a sensed property of the object O 1 ; and
(A) (6) mapping the first sensor data to the feature F 0 .
22. The non-transitory computer-readable medium of claim 1 , wherein (A) further comprises:
(A) (5) receiving, from a first sensor, first sensor data representing a sensed property of the object O 1 ; and
(A) (6) mapping the first sensor data to the feature F 0 .
23. The method of claim 21 , wherein the first sensor comprises a first imaging sensor, and wherein the first sensor data represents an optical input from the object O 1 .
24. A non-transitory computer-readable medium comprising computer program instructions executable by at least one computer processor to perform a method, the method comprising:
(A) generating, for each feature F in a plurality of features, a plurality of frequencies of observation of feature F in an object O 1 , wherein the plurality of features includes at least one physical feature and at least one use-based feature, comprising:
(A) (1) receiving first textual input from a first human;
(A) (2) receiving second textual input from a second human, wherein the first textual input differs from the second textual input;
(A) (3) mapping the first textual input and the second textual input to the same feature F 0 in the plurality of features; and
(A) (4) determining that the first textual input and the second textual input indicate that the object O 1 has feature F 0 ;
(B) generating output representing the plurality of frequencies of observation of each feature F in object O 1 ;
(C) identifying, based on the plurality of frequencies of observation of each feature F in object O 1 , a first subset of the plurality of features having frequencies satisfying a low frequency criterion, comprising:
(C) (1) generating, for each feature F in the plurality of features, a frequency count for feature F in object O 1 based on the plurality of frequencies of observation of feature F in object O 1 ; and
(C) (2) determining, for each feature F in the plurality of features, whether the frequency count for feature F satisfies the low frequency criterion, comprising:
determining that the feature count for a first one of the plurality of features satisfies the low frequency criterion; and
determining that the feature count for a second one of the plurality of features does not satisfy the low frequency criterion;
(D) automatically learning a first association between the first textual input and the feature F 0 , and storing first association data representing the first association;
(E) automatically learning a second association between the second textual input and the feature F 0 , and storing second association data representing the second association; and
wherein the frequency count for at least one feature F is equal to zero.
25. The non-transitory computer-readable medium of claim 24 , wherein (A) comprises:
(A) (1) generating, for each feature F in the plurality of features, a first indication of whether object O 1 was observed to have feature F, thereby generating a first plurality of indications for object O 1 ;
(A) (2) generating, for each feature F in the plurality of features, a second indication of whether object O 1 was observed to have feature F; thereby generating a second plurality of indications for object O 1 ; and
(A) (3) generating the plurality of frequencies of observation of feature F in object O 1 based on the first and second pluralities of indications for object O 1 .
26. The non-transitory computer-readable medium of claim 24 , wherein the output representing the plurality of frequencies of observation of feature F in object O 1 comprises a chart representing the plurality of frequencies of observation of feature F in object O 1 .
27. The non-transitory computer-readable medium of claim 26 , wherein the chart comprises a bar chart.
28. The non-transitory computer-readable medium of claim 26 , wherein the chart comprises a pie chart.
29. The non-transitory computer-readable medium of claim 24 , wherein the low frequency criterion comprises a maximum value, wherein (C) (2) comprises determining, for each feature F in the plurality of features, whether the frequency count for feature F is less than the maximum value, and wherein the first subset comprises features in the plurality of features having frequencies less than the maximum value.
30. The non-transitory computer-readable medium of claim 24 , wherein the method further comprises:
(F) identifying, based on the plurality of frequencies of observation of feature F in object O 1 , a second subset of the plurality of features having frequencies satisfying a high frequency criterion.
31. The non-transitory computer-readable medium of claim 30 , wherein the high frequency criterion comprises a minimum value, and wherein the second subset comprises features in the plurality of features having frequencies greater than the minimum value.
32. The non-transitory computer-readable medium of claim 24 , wherein the method further comprises:
(F) generating output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion.
33. The non-transitory computer-readable medium of claim 32 , wherein the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion includes output representing the frequencies satisfying the low frequency criterion.
34. The non-transitory computer-readable medium of claim 32 , wherein the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion comprises a chart.
35. The non-transitory computer-readable medium of claim 34 , wherein the chart comprises a bar chart.
36. The non-transitory computer-readable medium of claim 34 , wherein the chart comprises a pie chart.
37. The non-transitory computer-readable medium of claim 32 , wherein the output representing the plurality of frequencies of observation of feature F in object 0 1 includes the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion.
38. The non-transitory computer-readable medium of claim 37 , wherein the output representing the first subset of the plurality of features having frequencies satisfying the low frequency criterion comprises output emphasizing the first subset of the plurality of features.
39. The non-transitory computer-readable medium of claim 24 , wherein the object O is a physical object.
40. The non-transitory computer-readable medium of claim 24 :
wherein (D) comprises using a machine learning engine to automatically learn the first association; and
wherein (E) comprises using the machine learning engine to automatically learn the second association.
41. The non-transitory computer-readable medium of claim 24 , further comprising:
(F) receiving third textual input from the first human;
(G) receiving fourth textual input from the first human; and
(H) determining, based on at least one of the first association and the second association, that the third textual input and the fourth textual input indicate that a second object O 2 has feature F 0 .
42. The non-transitory computer-readable medium of claim 24 , wherein (H) comprises determining, based on the first association and the second association, that the third textual input and the fourth textual input indicate that a second object O 2 has feature F 0 .
43. The non-transitory computer-readable medium of claim 24 , wherein the first association data and the second association data are the same data.
44. The method of claim 21 , wherein the first sensor comprises a first imaging sensor, and wherein the first sensor data represents an optical input from the object O 1 .