DEVICES AND METHOD FOR SCORING DATA TO QUICKLY IDENTIFY RELEVANT ATTRIBUTES FOR INSTANT CLASSIFICATION

• US 20160019285A1
• Filed: 07/18/2014
• Published: 01/21/2016
• Est. Priority Date: 07/18/2014
• Status: Active Grant

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: Whelan, John Desmond
• Granted Patent: US 10,592,781 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 18/24 Classification techniques

Abstract

A method for allowing a computer to classify an input containing data. A list of categories is received. A sub list of categories is selected, wherein the sub-list comprises those categories in the list that have corresponding distinct correlation scores above a predetermined value. Input data that tends to over correlate to the classification system is received. A truncated snapshot is generated, the truncated snapshot comprising only attributes from the plurality of input attributes that have corresponding input categories that match categories in the sub-list of categories. The data is classified using the truncated snapshot and the classification system.

Claims

1. A method for allowing a computer to classify an input containing data, the method comprising:

receiving, by the computer, a list of categories from a classification system stored at the computer, wherein each category in the list is assigned a corresponding distinct correlation score that indicates a relevance of a particular category to a particular classification;

selecting, by the computer, a sub-list of categories, wherein the sub-list comprises those categories in the list that have corresponding distinct correlation scores above a predetermined value, and wherein the sub-list of categories comprises less than all of the categories in the list;

receiving, by the computer, the input containing the data, wherein the data is organized by a plurality of input attributes, wherein each input attribute has a corresponding input category and a corresponding input value, such that the plurality of input attributes has a plurality of input categories and a corresponding plurality of input values, wherein at least some of the plurality of input attributes has corresponding input categories that match at least some of a plurality of categories of the list, and wherein the plurality of input attributes tends to over-correlate with respect to the classification system, the tendency to over-correlate occurring as a result of an amount of the data;

generating, by the computer, a truncated snapshot, the truncated snapshot comprising only attributes from the plurality of input attributes that have the corresponding input categories that match categories in the sub-list of categories; and

classifying the data, by the computer, using the truncated snapshot and the classification system.

2. The method of claim 1 further comprising:

selecting, by the computer, the predetermined value by subtracting a first correlation score from a second correlation score to form a difference, and selecting the second correlation score as the predetermined value when the difference exceeds a given value.

3. The method of claim 2, wherein the predetermined value is selected by comparing the difference to other differences in correlation between adjacent categories higher on the list, and wherein the predetermined value selected when the difference exceeds a given percentage change in correlation relative to the other differences in correlation.

4. The method of claim 1 further comprising:

prior to receiving the data, training the classification system by using known data collected from known observations and instructing the classification system as to which of the known observations falls into a given classification of the classification system.

5. The method of claim 1, wherein the computer comprises a processor and an associative memory, the associative memory comprising a plurality of data and a plurality of associations among the plurality of data, wherein the plurality of data is collected into associated groups, wherein the associative memory is configured to be queried based on at least indirect relationships among the plurality of data, wherein the data is received in the associative memory, and wherein the step of selecting the sub-list further comprises:

the associative memory cycling through all of the plurality of categories, determining whether a given category has a corresponding correlation score above the predetermined value, and including the given category in the sub-list if the corresponding correlation score is above the predetermined value.

6. The method of claim 1, wherein the data comprises a snapshot, the snapshot defined as an instrument used to initiate a response from the classification system given a new observation, wherein the snapshot is generated outside of the classification system using data values unknown to the classification system.

7. The method of claim 1, wherein the list comprises an ordered list, ordered from highest corresponding distinct correlation score to lowest corresponding distinct correlation score.

8. The method of claim 1, wherein selecting the sub-list is performed at a time selected from the group consisting of:

before receiving the input and after receiving the input.

9. A data processing system comprising:

a processor;

a bus connected to the processor;

a non-transitory computer readable storage medium connected to the bus, the non-transitory computer readable storage medium storing a computer program product which, when executed by the processor, performs a computer implemented method for allowing a computer to classify an input containing data, the computer program product comprising;

computer usable program code for receiving a list of categories from a classification system stored at the computer, wherein each category in the list is assigned a corresponding distinct correlation score that indicates a relevance of a particular category to a particular classification;

computer usable program code for selecting a sub-list of categories, wherein the sub-list comprises those categories in the list that have corresponding distinct correlation scores above a predetermined value, and wherein the sub-list of categories comprises less than all of the categories in the list;

computer usable program code for receiving the input containing the data, wherein the data is organized by a plurality of input attributes, wherein each input attribute has a corresponding input category and a corresponding input value, such that the plurality of input attributes has a plurality of input categories and a corresponding plurality of input values, wherein at least some of the plurality of input attributes has corresponding input categories that match at least some of a plurality of categories of the list of categories, and wherein the plurality of input attributes tends to over-correlate with respect to the classification system, the tendency to over-correlate occurring as a result of an amount of the data;

computer usable program code for generating a truncated snapshot, the truncated snapshot comprising only attributes from the plurality of input attributes that have corresponding input categories that match categories in the sub-list of categories; and

computer usable program code for classifying the data using the truncated snapshot and the classification system.

10. The data processing system of claim 9, wherein the computer program product further comprises:

computer usable program code for selecting the predetermined value by subtracting a first correlation score from a second correlation score to form a difference, and selecting the second correlation score as the predetermined value when the difference exceeds a given value.

11. The data processing system of claim 10, wherein the predetermined value is selected by comparing the difference to other differences in correlation between adjacent categories higher on the list, and wherein the predetermined value is selected when the difference exceeds a given percentage change in correlation relative to the other differences in correlation.

12. The data processing system of claim 9, wherein the computer program product further comprises:

computer usable program code for, prior to receiving the data, training the classification system by using known data collected from known observations and instructing the classification system as to which of the known observations falls into a given classification of the classification system.

13. The data processing system of claim 9, wherein the data comprises a snapshot, the snapshot defined as an instrument used to initiate a response from the classification system given a new observation, wherein the snapshot is generated outside of the classification system using data values unknown to the classification system.

14. A system configured to classify an input containing data, the system comprising:

a tangible input device configured to receive a list of categories from a classification system stored at a computer, wherein each category in the list is assigned a corresponding distinct correlation score that indicates a relevance of a particular category to a particular classification;

a selection device configured to select a sub-list of categories, wherein the sub-list comprises those categories in the list that have corresponding distinct correlation scores above a predetermined value, and wherein the sub-list of categories comprises less than all of the categories in the list;

a data input device configured to receive the input, the input containing data, wherein the data is organized by a plurality of input attributes, wherein each input attribute has a corresponding input category and a corresponding input value, such that the plurality of input attributes has a plurality of input categories and a corresponding plurality of input values, wherein at least some of the plurality of input attributes has corresponding input categories that match at least some of a plurality of categories of the list, and wherein the plurality of input attributes tends to over-correlate with respect to the classification system, the tendency to over-correlate occurring as a result of an amount of the data;

a snapshot generator configured to generate a truncated snapshot, the truncated snapshot comprising only attributes from the plurality of input attributes that have corresponding input categories that match categories in the sub-list of categories; and

a classifier configured to classify the data using the truncated snapshot and the classification system.

15. The system of claim 14, wherein the selection device is further configured to select the predetermined value by subtracting a first correlation score from a second correlation score to form a difference, and to select the second correlation score as the predetermined value when the difference exceeds a given value.

16. The system of claim 15, wherein the selection device is further configured to compare the difference to other differences in correlation between adjacent categories higher on the list, and wherein the predetermined value is selected when the difference exceeds a given percentage change in correlation relative to the other differences in correlation.

17. The system of claim 14 further comprising:

a trainer configured to, prior to receiving the data, train the classification system by using known data collected from known observations and to instruct the classification system as to which of the known observations falls into a given classification of the classification system.

18. The system of claim 14, wherein the system further comprises:

an associative memory, the associative memory comprising a plurality of data and a plurality of associations among the plurality of data, wherein the plurality of data is collected into associated groups, wherein the associative memory is configured to be queried based on at least indirect relationships among the plurality of data, wherein the data is received in the associative memory, and wherein the selection device is configured to select the sub-list by;

the associative memory cycling through all of the plurality of categories, determining whether a given category has a corresponding correlation score above the predetermined value, and including the given category in the sub-list if the corresponding correlation score is above the predetermined value.

19. The system of claim 14, wherein the data comprises a snapshot, the snapshot defined as an instrument used to initiate a response from the classification system given a new observation, wherein the snapshot is generated outside of the classification system using data values unknown to the classification system.

20. The system of claim 14, wherein the list comprises an ordered list, ordered from highest corresponding distinct correlation score to lowest corresponding distinct correlation score.