METHODS AND APPARATUS TO IMPROVE FEATURE ENGINEERING EFFICIENCY WITH METADATA UNIT OPERATIONS

• US 20180089424A1
• Filed: 09/29/2016
• Published: 03/29/2018
• Est. Priority Date: 09/29/2016
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Yang, Chih-Yuan, Gai, Yi
• Granted Patent: US 10,607,004 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 21/52 during program execution, e.g. stack integrity ; Preventing unwanted data erasure; Buffer overflow
• CPC Class Codes: G06F 21/552 involving long-term monitoring or reporting
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]

Abstract

Methods, apparatus, systems and articles of manufacture are disclosed to improve feature engineering efficiency. An example method disclosed herein includes retrieving a log file in a first file format, the log file containing feature occurrence data, generating a first unit operation based on the first file format to extract the feature occurrence data from the log file to a string, the first unit operation associated with a first metadata tag, generating second unit operations to identify respective features from the feature occurrence data, the second unit operations associated with respective second metadata tags, and generating a first sequence of the first metadata tag and the second metadata tags to create a first vector output file of the feature occurrence data

Claims

1. A computer-implemented method to apply feature engineering with metadata-driven unit operations, comprising:

retrieving a log file in a first file format, the log file containing feature occurrence data;

generating a first unit operation based on the first file format to extract the feature occurrence data from the log file to a string, the first unit operation associated with a first metadata tag;

generating second unit operations to identify respective features from the feature occurrence data, the second unit operations associated with respective second metadata tags; and

generating a first sequence of the first metadata tag and the second metadata tags to create a first vector output file of the feature occurrence data.

2. The computer-implemented method as defined in claim 1, wherein the first unit operation includes parsing operations for at least one of a text file format, a comma separated value (CSV) file format, a JavaScript Object Notation (JSON) file format, or a binary file format.

3. The computer-implemented method as defined in claim 1, further including building a dictionary of feature nomenclature associated with the respective features from the feature occurrence data.

4. The computer-implemented method as defined in claim 3, further including generating search substrings of the feature nomenclature.

5. The computer-implemented method as defined in claim 4, wherein respective ones of the second unit operations identify the search substrings as feature occurrence instances.

6. The computer-implemented method as defined in claim 1, further including:

executing the first sequence of the first metadata tag and the second metadata tags to create the first vector output file associated with the first file format; and

executing a second sequence of the first metadata tag and alternate ones of the second metadata tags to create a second vector output file associated with the first file format.

7. The computer-implemented method as defined in claim 6, wherein the second metadata tags invoke a dictionary to identify feature occurrence instances, and the alternate ones of the second metadata tags invoke regular expressions to identify feature occurrence instances.

8. (canceled)

9. (canceled)

10. (canceled)

11. (canceled)

12. An apparatus to apply feature engineering with metadata-driven unit operations, comprising:

a log file retriever to retrieve a log file in a first file format, the log file containing feature occurrence data;

a file to string operation builder to generate a first unit operation based on the first file format to extract the feature occurrence data from the log file to a string, the first unit operation associated with a first metadata tag;

an extraction operation builder to generate second unit operations to identify respective features from the feature occurrence data, the second unit operations associated with respective second metadata tags; and

an operation flow builder to generate a first sequence of the first metadata tag and the second metadata tags to create a first vector output file of the feature occurrence data.

13. The apparatus as defined in claim 12, wherein the file to string operation builder is to generate parsing operations for at least one of a text file format, a comma separated value (CSV) file format, a JavaScript Object Notation (JSON) file format, or a binary file format.

14. The apparatus as defined in claim 12, further including a dictionary editor to build a dictionary of feature nomenclature associated with the respective features from the feature occurrence data.

19. The apparatus as defined in claim 12, further including a dictionary editor to facilitate extraction of feature instances from respective ones of the second unit operations based on at least one of dictionary matching or regular expression strings.

20. The apparatus as defined in claim 12, further including a dictionary editor to normalize respective ones of the second unit operations to identify feature nomenclature based on a dictionary association.

21. The apparatus as defined in claim 12, further including a hashing operation builder to hash respective ones of the second unit operations to a unique integer value.

22. The apparatus as defined in claim 12, further including a formatting operation builder to format the vector output file based on a Library for Support Vector Machines (LIBSVM) classification format.

15. (canceled)

16. (canceled)

17. (canceled)

18. (canceled)

23. A tangible computer readable storage medium comprising computer readable instructions which, when executed, cause a processor to at least:

retrieve a log file in a first file format, the log file containing feature occurrence data;

generate a first unit operation based on the first file format to extract the feature occurrence data from the log file to a string, the first unit operation associated with a first metadata tag;

generate second unit operations to identify respective features from the feature occurrence data, the second unit operations associated with respective second metadata tags; and

generate a first sequence of the first metadata tag and the second metadata tags to create a first vector output file of the feature occurrence data.

24. The computer readable storage medium of claim 23, wherein the instructions, when executed, cause the processor to generate parsing operations for at least one of a text file format, a comma separated value (CSV) file format, a JavaScript Object Notation (JSON) file format, or a binary file format.

25. The computer readable storage medium of claim 23, wherein the instructions, when executed, cause the processor to build a dictionary of feature nomenclature associated with the respective features from the feature occurrence data.

26. The computer readable storage medium of claim 25, wherein the instructions, when executed, cause the processor to generate search substrings of the feature nomenclature.

27. The computer readable storage medium of claim 26, wherein the instructions, when executed, cause the processor to identify, from respective ones of the second unit operations, the search substrings as feature occurrence instances.

28. The computer readable storage medium of claim 23, wherein the instructions, when executed, cause the processor to:

execute the first sequence of the first metadata tag and the second metadata tags to create the first vector output file associated with the first file format; and

execute a second sequence of the first metadata tag and alternate ones of the second metadata tags to create a second vector output file associated with the first file format.

29. (canceled)

30. (canceled)

31. (canceled)

32. (canceled)

33. (canceled)