Defect predicate expression extraction

• US 8,484,622 B2
• Filed: 04/15/2011
• Issued: 07/09/2013
• Est. Priority Date: 04/27/2010
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Kuboki, Mutsugu, Kurokawa, Yuki, Marakami, Takuma, Nasukawa, Tetsuya, Unno, Yuya
• Primary Examiner(s): Dao, Thuy
• Assistant Examiner(s): BERMAN, STEPHEN DAVID
• Application Number:
  US13/087,639
• Publication Number:
  US 20110265065A1
• Time in Patent Office: 816 Days
• Field of Search: None
• US Class Current: 717/131
• CPC Class Codes: G06F 11/008 Reliability or availability analysis

Abstract

A defect predicate expression extraction device. The device extracts, as candidates for predicate expressions representing defects, predicate expressions occurring in the neighborhood of predicate modifying expressions representing suddenness or predicate modifying expressions representing repeatability. The defect predicate expression extraction device further extracts, as predicate expressions representing normality, predicate expressions occurring in the neighborhood of predicate modifying expressions representing normality and extracts predicate expressions representing defects by removing the predicate expressions representing normality from a list of the candidates for predicate expressions representing defects.

Claims

1. A defect predicate expression extraction method executed by a processor device for extracting predicate expressions representing defects from text data related to use of products, the method comprising:

storing in a first predicate modifying expression storage unit, as predicate modifying expressions for detecting defect expressions, at least one of predicate modifying expressions representing suddenness and predicate modifying expressions representing repeatability;

detecting, in said text data, expressions matching each of the said stored predicate modifying expressions for detecting defect expressions and extracting, as a candidate for a predicate expression representing a defect, a predicate expression occurring in a neighborhood of each of said detected expressions matching each of the said stored predicate modifying expressions for detecting defect expressions in said text data;

storing in a second predicate modifying expression storage unit, as predicate modifying expressions for detecting normality expressions, predicate modifying expressions representing normality;

detecting, in said text data, expressions matching each of the said stored predicate modifying expressions for detecting normality expressions and extracting, as a predicate expression representing normality, a predicate expression occurring in a neighborhood of each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting normality expressions in said text data; and

acquiring, as said predicate expressions representing defects, predicate expressions remaining after said extracted predicate expressions are removed from a list of said predicate expressions extracted as said candidates for said predicate expressions representing defects.

2. The defect predicate expression extraction method according to claim 1,

wherein said predicate modifying expressions representing suddenness include at least one expression selected from the group consisting of:

a predetermined number of predicate modifying expressions representing suddenness;

wherein said predicate modifying expressions representing repeatability include at least one expression selected from the group consisting of;

a predetermined number of predicate modifying expressions representing repeatability; and

wherein said predicate modifying expressions representing normality include at least one expression selected from the group consisting of;

a predetermined number of predicate modifying expressions representing normality.

3. The defect predicate expression extraction method according to claim 1,

wherein extracting a candidate for a predicate expression representing a defect further comprises extracting a predicate expression directly modified by each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting defect expressions;

and

wherein extracting a predicate expression representing normality further comprises extracting a predicate expression directly modified by each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting normality expressions.

4. The defect predicate expression extraction method according to claim 1,

wherein extracting a candidate for a predicate expression representing a defect further comprises counting a number of occurrences of said predicate expression occurring in said neighborhood of at least one of the said predicate modifying expressions stored in said first predicate modifying expression storage unit;

wherein extracting a predicate expression representing normality further comprises counting a number of occurrences of said predicate expression occurring in a neighborhood of at least one of the said predicate modifying expressions stored in said second predicate modifying expression storage unit; and

wherein a same predicate expression is extracted as both representing a defect and representing normality, determining, on the basis of the respective numbers of occurrences of said counted predicate expression, whether to set said same predicate expression to be said predicate expression representing a defect.

5. The defect predicate expression extraction method according to claim 1, further comprising:

extracting, from said text data, predicate modifying expressions co-occurring with a predicate expression representing a specific defect; and

registering said predicate modifying expressions in said first predicate modifying expression storage unit.

6. The defect predicate expression extraction method according to claim 1, further comprising:

storing said predicate expressions representing defects in a defect predicate expression storage unit;

detecting, in said text data, expressions matching each of the said stored predicate expressions representing defects and extracting a noun expression occurring in a neighborhood of each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting normality expressions in said text data; and

storing, in an analysis object storage unit, as an object to be analyzed, a pair of said detected expressions matching said predicate expression representing said defect and said extracted noun expression, in association with a frequency of extraction of said object to be analyzed.

7. The defect predicate expression extraction method according to claim 6, further comprising:

calculating a correlation value of each of the said stored objects to be analyzed; and

generating said correlation value of said object to be analyzed as an analysis result.

8. A defect predicate expression extraction device including a memory and a processor device configured to extract predicate expressions representing defects from text data related to use of products, comprising:

a predicate modifying expression storage unit for detecting defect expressions, wherein said predicate modifying expression storage unit stores at least one of predicate modifying expressions representing suddenness and predicate modifying expressions representing repeatability;

a defect predicate expression candidate extraction unit for detecting, in said text data, expressions matching each of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting defect expressions, and extracting, as a candidate for a predicate expression representing a defect, a predicate expression occurring in a neighborhood of each of said detected expressions matching each of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting defect expressions in said text data;

a predicate modifying expression storage unit for detecting normality expressions, wherein said predicate modifying expression storage unit stores predicate modifying expressions representing normality;

a normality predicate expression extraction unit for detecting, in said text data, expressions matching each of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting normality expressions, and extracting, as a predicate expression representing normality, a predicate expression occurring in a neighborhood of each of the said detected expressions matching each of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting normality expressions in said text data; and

a defect predicate expression acquisition unit for acquiring, as said predicate expressions representing defects, predicate expressions remaining after said predicate expressions extracted by said normality predicate expression extraction unit are removed from a list of said predicate expressions extracted as candidates for said predicate expressions representing defects.

9. The defect predicate expression extraction device according to claim 8,

wherein said predicate modifying expressions representing suddenness include at least one expression selected from the group consisting of:

a predetermined number of predicate modifying expressions representing suddenness;

wherein said predicate modifying expressions representing repeatability include at least one expression selected from the group consisting of;

a predetermined number of predicate modifying expressions representing repeatability; and

wherein said predicate modifying expressions representing normality include at least one expression selected from the group consisting of;

a predetermined number of predicate modifying expressions representing normality.

10. The defect predicate expression extraction device according to claim 8,

wherein said defect predicate expression candidate extraction unit further extracts, as a candidate for said predicate expression representing said defect, a predicate expression directly modified by each of the said detected expressions matching each of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting defect expressions;

and

wherein said normality predicate expression extraction unit further extracts, as a predicate expression representing normality, a predicate expression directly modified by each of the said detected expressions matching each of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting normality expressions.

11. The defect predicate expression extraction device according to claim 8,

wherein said defect predicate expression candidate extraction unit counts a number of occurrences of said predicate expression occurring in a neighborhood of at least one of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting defect expressions;

wherein said normality predicate expression extraction unit counts a number of occurrences of said predicate expression occurring in a neighborhood of at least one of the said predicate modifying expressions stored in said predicate modifying expression storage unit for detecting normality expressions; and

wherein in a case where a same predicate expression is extracted by both the said defect predicate expression candidate extraction unit and the said normality predicate expression extraction unit, said defect predicate expression acquisition unit determines, on the basis of the respective numbers of occurrences of said predicate expression counted by said defect predicate expression candidate extraction unit and said normality predicate expression extraction unit, whether to set the same predicate expression to be said predicate expression representing a defect.

12. The defect predicate expression extraction device according to claim 8, further comprising a predicate modifying expression registration unit for detecting defect expressions, said predicate modifying expression registration unit extracting, from text data related to use of products, predicate modifying expressions co-occurring with a predicate expression representing a specific defect and registering said predicate modifying expressions in said predicate modifying expression storage unit for detecting defect expressions.

13. The defect predicate expression extraction device according to claim 8, further comprising:

a defect predicate expression storage unit for storing said predicate expressions representing defects acquired by said defect predicate expression acquisition unit;

an analysis object extraction unit for detecting, in said text data to be analyzed, expressions matching each of the said predicate expressions representing defects stored in said defect predicate expression storage unit and extracting a noun expression occurring in a neighborhood of each of the said detected expressions matching each of the said predicate expressions representing defects stored in said defect predicate expression storage unit in said text data to be analyzed; and

an analysis object storage unit for storing, as an object to be analyzed, a pair of said detected expressions matching said predicate expression representing said defect and said extracted noun expression, in association with a frequency of extraction of said object to be analyzed.

14. The defect predicate expression extraction device according to claim 13, further comprising an analysis result generation unit for calculating a correlation value of each of the said objects to be analyzed stored in said analysis object storage unit and generating said correlation value of said object to be analyzed as an analysis result.

15. A computer program product having a non-transitory computer readable medium tangibly embodying computer readable instructions which, when executed, cause a computer to carry out the steps of a method for extracting predicate expressions representing defects from text data related to use of products, the method comprising:

storing in a first predicate modifying expression storage module, as predicate modifying expressions for detecting defect expressions, at least one of predicate modifying expressions representing suddenness and predicate modifying expressions representing repeatability;

detecting, in said text data, expressions matching each of the said stored predicate modifying expressions for detecting defect expressions and extracting, as a candidate for a predicate expression representing a defect, a predicate expression occurring in a neighborhood of each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting defect expressions in said text data;

storing in a second predicate modifying expression storage module, as predicate modifying expressions for detecting normality expressions, predicate modifying expressions representing normality;

detecting, in said text data, expressions matching each of the said stored predicate modifying expressions for detecting normality expressions and extracting, as a predicate expression representing normality, a predicate expression occurring in a neighborhood of each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting normality expressions in said text data; and

acquiring, as said predicate expressions representing defects, predicate expressions remaining after said extracted predicate expressions are removed from a list of said predicate expressions extracted as said candidates for said predicate expressions representing defects.

16. The computer program product according to claim 15,

wherein said predicate modifying expressions representing suddenness include at least one expression selected from the group consisting of:

a predetermined number of predicate modifying expressions representing suddenness;

wherein said predicate modifying expressions representing repeatability include at least one expression selected from the group consisting of;

a predetermined number of predicate modifying expressions representing repeatability; and

wherein said predicate modifying expressions representing normality include at least one expression selected from the group consisting of;

a predetermined number of predicate modifying expressions representing normality.

17. The computer program product according to claim 15,

wherein extracting a candidate for a predicate expression representing a defect further comprises extracting a predicate expression directly modified by each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting defect expressions;

and

wherein extracting a predicate expression representing normality further comprises extracting a predicate expression directly modified by each of the said detected expressions matching each of the said stored predicate modifying expressions for detecting normality expressions.

18. The computer program product according to claim 15,

wherein extracting a candidate for a predicate expression representing a defect further comprises counting a number of occurrences of said predicate expression occurring in said neighborhood of at least one of the said predicate modifying expressions stored in said first predicate modifying expression storage module;

wherein extracting a predicate expression representing normality further comprises counting a number of occurrences of said predicate expression occurring in a neighborhood of at least one of the said predicate modifying expressions stored in said second predicate modifying expression storage module; and

wherein a same predicate expression is extracted as both representing a defect and representing normality, determining, on the basis of the respective numbers of occurrences of said counted predicate expression, whether to set said same predicate expression to be said predicate expression representing a defect.

19. The computer program product according to claim 15, further comprising computer readable program code configured to perform the steps of:

extracting, from said text data, predicate modifying expressions co-occurring with a predicate expression representing a specific defect; and

registering said predicate modifying expressions in said first predicate modifying expression storage module.

20. The computer program product according to claim 15, further comprising computer readable program code configured to perform the steps of:

storing said predicate expressions representing defects in a defect predicate expression storage module;

detecting, in said text data, expressions matching each of the said stored predicate expressions representing defects and extracting a noun expression occurring in a neighborhood of each of the said detected expressions matching each of the said stored predicate expressions representing defects in said text data;

storing, in an analysis object storage module, as an object to be analyzed, a pair of said detected expressions matching said predicate expression representing said defect and said extracted noun expression, in association with a frequency of extraction of said object to be analyzed;

calculating a correlation value of each of the said stored objects to be analyzed; and

generating said correlation value of said object to be analyzed as an analysis result.