Data classifier system, data classifier method and data classifier program

• US 9,361,367 B2
• Filed: 07/29/2009
• Issued: 06/07/2016
• Est. Priority Date: 07/30/2008
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Mizuguchi, Hironori, Tateishi, Kenji, Hosomi, Itaru, Kusui, Dai
• Primary Examiner(s): KUDDUS, DANIEL A
• Application Number:
  US13/056,031
• Publication Number:
  US 20110179037A1
• Time in Patent Office: 2,505 Days
• Field of Search: 707/740, 707/797, 707/956
• US Class Current: 1/1
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/353 into predefined classes

Abstract

A data classifier system of the present invention selects a plurality of classifications correlated to data groups so as to output classification axes based on hierarchical classifications and data groups. The data classifier system includes a basic category accumulation means, a classification axis candidate creation means and a priority calculation means. The basic category accumulation means accumulates classifications serving as basic categories used for selecting desired classifications in advance. The classification axis candidate creation means creates classification axis candidates based on combinations of classifications each correlated to at least one data among descendant classifications of each basic category. The priority calculation means calculates priorities with respect to the classification axis candidates created by the classification axis candidate creation means based on hierarchical distances of classifications in the classified hierarchy.

Claims

1. A data classifier system, implemented by a computer including a processor and a storage device, which refers to a classified hierarchy including a plurality of classifications to produce a plurality of classification axes based on classifications and data groups input,

wherein the storage device stores the classified hierarchy establishing a parent-child relationship between the plurality of classifications and classifications serving as basic categories which are used to select desired classifications;

wherein the processor creates classification axis candidates based on combinations of classifications belonging to basic categories;

wherein the processor inputs the classification axis candidates to calculate priorities with respect to the classification axis candidates with reference to hierarchical distances, each representing a length of a path reaching a common ancestor among classifications associated with the classification axis candidates, and hierarchical depths, each representing a length of a path between classifications associated with the classification axis candidates, and

wherein semantic independence of each of the plurality of classification axes is based on length of hierarchical distances of each of the plurality of classification axes,

wherein the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy is calculated using a formula;

Priority(X;

C)=W1×

Independence(X;

C)+W2×

Specifics(X;

C)+W3×

Exhaustivity(X;

C)+W4×

Uniqueness(X;

C),

wherein W1, W2, W3 and W4 denote weight coefficients to indexes which are determined in advance,

wherein the Exhaustivity(X;

C) is given using a formula;

Exhaustivity(X;

C)=1/DataNum×

|U Data(ci)|,

wherein DataNum denotes an amount of data being classified,

wherein Data(ci) denotes a set of data allocated to a classification ci selected from among the plurality of classifications, and

wherein |U Data (ci)| denotes an amount of data allocated to the plurality of classifications.

2. The data classifier system according to claim 1, wherein the processor determines the priorities of the classification axis candidates based on at least one of the lengths of the paths reaching the common ancestor in the classified hierarchy, data quantities, each representing a number of subclassifications descendant from each classification, and data redundancies each representing a redundancy of classifications assigned to each classification axis.

3. The data classifier system according to claim 2, wherein each of the data quantities indicates one of (a) coverage and (b) exhaustivity relating to the plurality of classifications assigned to each classification axis, and wherein the processor increases the priority as the coverage increases.

4. The data classifier system according to claim 2, wherein the processor reduces the priority as the data redundancy increases.

5. The data classifier system according to claim 1, wherein the processor reduces a number of classification axis candidates based on data quantities of classifications and hierarchical distances among classifications in the classified hierarchy.

6. The data classifier system according to claim 1, wherein the processor selects classification axes with data quantities of classifications which are larger than a predetermined value or within an upper range in a descending order of data quantities, and within a specific range of hierarchical distances.

7. The data classifier system according to claim 1, wherein the processor displays the classification axis candidates and the priorities in association with data groups, and wherein the processor displays the classification axis candidates in an order of the priorities, selects the classification axis candidates to change and display classification axes, and selects classifications in the classification axes to limit data groups.

8. The data classifier system according to claim 1, wherein the processor retrieves data groups based on retrieval keywords, creates classification axis candidates based on the retrieved data groups, and calculates priorities of classification axes correlated to the retrieved data groups.

9. The data classifier system according to claim 1, wherein the processor establishes correlation between the classifications and the data groups input.

10. The data classifier system according to claim 1, wherein the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy is calculated using a formula:

Priority(X;

C)=1/Max(X)×

1/(2×

N)×

Σ

ComDist(ci,cj),

wherein Max(X) denotes a depth of a deepest classification among classifications descending from the basic category X,

N denotes a number of the plurality of classifications, and

ComDist(ci,cj) denotes a distance between classifications ci and cj selected from among the plurality of classifications.

11. The data classifier system according to claim 1, wherein the Independent(X:

C) is given using a formula;

Independence(X;

C)=1/Max(X)×

1/(2×

N)×

Σ

ComDist(ci,cj),

wherein Max(X) denotes a depth of a deepest classification among classifications descending from the basic category X,

N denotes a number of the plurality of classifications, and

ComDist(ci,cj) denotes a distance between classifications ci and cj selected from among the plurality of classifications.

12. The data classifier system according to claim 1, wherein the Specifics(X:

C) is given using a formula;

Specifics(X;

C)=1/Max(X)×

1/N×

Σ

Depth(X,ci),

wherein Max(X) denotes a depth of a deepest classification among classifications descending from the basic category,

N denotes a number of the plurality of classifications, and

Depth(X,ci) denotes a shortest path length from the basic category X to a classification ci selected from among the plurality of classifications.

29. The data classifier system according to claim 1, wherein the processor calculates the priority by dividing an average value of the hierarchical distances, among classifications in the classified hierarchy, by the hierarchical depth.

13. A data classifier system, implemented by a computer including a processor and a storage device, which refers to a classified hierarchy including a plurality of classifications to create a plurality of classification axes based on classifications and data groups input,

wherein the storage device stores a classified hierarchy establishing a parent-child relationship between the plurality of classifications and classifications serving as basic categories which are used to select desired classifications;

wherein the processor creates classification axis candidates based on combinations of classifications belonging to each basic category to produce multidimensional classification axis candidates each combining a plurality of classification axis candidates;

wherein the processor inputs the classification axis candidates to calculate-priorities with respect to the multidimensional classification axis candidates with reference to hierarchical distances, each representing a length of a path reaching a common ancestor among classifications associated with classification axis candidates and hierarchical depths each representing a length of a path between classifications associated with classification axis candidates, and

wherein semantic independence of each of the plurality of classification axes is based on length of hierarchical distances of each of the plurality of classification axes,

wherein the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy is calculated using a formula;

Priority(X;

C)=W1×

Independence(X;

C)+W2×

Specifics(X;

C)+W3×

Exhaustivity(X;

C)+W4×

Uniqueness(X;

C),

wherein W1, W2, W3 and W4 denote weight coefficients to indexes which are determined in advance,

wherein the Uniqueness(X;

C) is given using a formula;

Uniqueness(X;

C)=1/(1/|U Data(ci)|×

Σ

CatNum(ci)),

wherein Data(ci) denotes a set of data allocated to a classification ci selected from among the plurality of classifications,

wherein |U Data (ci)| denotes an amount of data having no redundancy allocated to each of the plurality of classifications, and

wherein CatNum(ci) denotes an amount of data allocated to the classification ci.

14. The data classifier system according to claim 13, wherein the processor determines the priorities of the multidimensional classification axis candidates based on at least one of the lengths of the paths reaching the common ancestor among classifications in the classified hierarchy, data quantities, each representing a number of subclassifications descendant from each classification, and data redundancies each representing a redundancy between classifications assigned to each classification axis.

15. The data classifier system according to claim 14, wherein each of the data quantities indicates one of (a) coverage and (b) exhaustivity relating to the plurality of classifications assigned to each classification axis, and wherein the processor increases the priority as the coverage increases.

16. The data classifier system according to claim 14, wherein the processor reduces the priority as the data redundancy increases.

17. The data classifier system according to claim 13, wherein the processor reduces a number of multidimensional classification axis candidates based on data quantities of classifications of classification axes for each dimension and hierarchical distances between classifications of classification axes for each dimension.

18. The data classifier system according to claim 13, wherein the processor selects multidimensional classification axes including classifications with data quantities of classifications of classification axes for each dimension which are larger than a predetermined value or within an upper range in a descending order of data quantities, and classifications within a specific range of hierarchical distances among classifications of classification axes for each dimension.

19. The data classifier system according to claim 13,

wherein the processor displays the multidimensional classification axis candidates and the priorities in association with data groups in a list form or in a table form;

and

wherein the processor selects the multidimensional classification axis candidates to display and allocate classifications for each dimension to the list form or the table form, and subsequently selects one or plural classifications to display at least one of data quantities, data names, data attributes and characteristic terms with respect to the selected classifications.

20. The data classifier system according to claim 13, wherein the processor retrieves data groups based on retrieval keywords to output retrieved data groups.

21. The data classifier system according to claim 13, wherein the processor establishes correlation between the classifications and the data groups input.

22. The data classifier system according to claim 13, wherein the processor calculates the priority by dividing an average value of the hierarchical distances, among classifications in the classified hierarchy, by the hierarchical depth.

23. A data classifier method, implemented by a computer including a processor and a memory, which refers to a classified hierarchy including a plurality of classifications to produce a plurality of classification axes based on classifications and data groups input,

wherein the memory stores the classified hierarchy establishing a parent-child relationship between the plurality of classifications and classifications serving as basic categories which are used to select desired classifications;

wherein the processor creates classification axis candidates based on combinations of classifications belonging to each basic category;

wherein the processor inputs the classification axis candidates to calculate priorities with respect to the classification axis candidates with reference to hierarchical distances, each representing a length of a path reaching a common ancestor among classifications associated with classification axis candidates, and hierarchical depths each representing a length of a path between classifications associated with the classification axis candidates, and

wherein semantic independence of each of the plurality of classification axes is based on length of hierarchical distances of each of the plurality of classification axes,

wherein the processor calculates the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy using a formula;

Priority(X;

C)=W1×

Independence(X;

C)+W2×

Specifics(X;

C)+W3×

Exhaustivity(X;

C)+W4×

Uniqueness(X;

C),

wherein W1, W2, W3 and W4 denote weight coefficients to indexes which are determined in advance,

wherein the Exhaustivity(X;

C) is given using a formula;

Exhaustivity(X;

C)=1/DataNum×

|U Data(ci)|,

wherein DataNum denotes an amount of data being classified,

wherein Data(ci) denotes a set of data allocated to a classification ci selected from among the plurality of classifications, and

wherein |U Data (ci)| denotes an amount of data allocated to the plurality of classifications.

24. The data classifier method according to claim 23, wherein the processor determines the priorities of the classification axis candidates based on at least one of the hierarchical distances between classifications in the classified hierarchy, data quantities, representing a number of subclassifications descendant from each classification, and data redundancies each representing a redundancy of classifications assigned to each classification axis.

25. A data classifier method, implemented by a computer including a processor and a memory, which refers to a classified hierarchy including a plurality of classifications to produce a plurality of classification axes based on classifications and data groups input,

wherein the memory stores the classified hierarchy establishing a parent-child relationship between the plurality of classifications and classifications serving as basic categories which are used to select desired classifications, wherein the processor creates classification axis candidates based on combinations of classifications belonging to each basic category, to produce multidimensional classification axis candidates each combining a plurality of classification axis candidates;

wherein the processor inputs the multidimensional classification axis candidates to calculate priorities with respect to the multidimensional classification axis candidates with reference to the hierarchical distances, each representing a length of a path reaching a common ancestor among classifications associated with classification axis candidates, and hierarchical depths each representing a length of a path between classifications associated with classification axis candidates; and

wherein semantic independence of each of the plurality of classification axes is based on length of hierarchical distances of each of the plurality of classification axes,

wherein the processor calculates the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy using a formula;

Priority(X;

C)=W1×

Independence(X;

C)+W2×

Specifics(X;

C)+W3×

Exhaustivity(X;

C)+W4×

Uniqueness(X;

C),

wherein W1, W2, W3 and W4 denote weight coefficients to indexes which are determined in advance,

wherein the Uniqueness(X;

C) is given using a formula;

Uniqueness(X;

C)=1/(1/|U Data(ci)|×

Σ

CatNum(ci)),

wherein Data(ci) denotes a set of data allocated to a classification ci selected from among the plurality of classifications,

wherein |U Data (ci)| denotes an amount of data having no redundancy allocated to each of the plurality of classifications, and

wherein CatNum(ci) denotes an amount of data allocated to the classification ci.

26. The data classifier method according to claim 25, wherein the processor determines the priorities of the multidimensional classification axis candidates based on at least one of the hierarchical distances between classifications in the classified hierarchy, data quantities of classifications and data redundancies of classifications.

27. A non-transitory computer-readable storage medium storing a data classifier program, executable by a computer, which refers to a classified hierarchy including a plurality of classifications to produce a plurality of classification axes based on classifications and data groups input, said data classifier program implementing:

a basic category accumulation process which stores the classified hierarchy establishing a parent-child relationship between the plurality of classifications and classifications serving as basic categories which are used to select desired classifications;

a classification axis candidate creation process which creates classification axis candidates based on combinations of classifications belonging to each basic category; and

a priority calculation process which inputs the classification axis candidates to calculate priorities with respect to the classification axis candidates with reference to hierarchical distances, each representing a length of a path reaching a common ancestor among classifications associated with the classification axis candidates, and hierarchical depths, each representing a length of a path between classifications associated with the classification axis candidates,

wherein semantic independence of each of the plurality of classification axes is based on length of hierarchical distances of each of the plurality of classification axes,

wherein the priority calculation process calculates the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy using a formula;

Priority(X;

C)=W1×

Independence(X;

C)+W2×

Specifics(X;

C)+W3×

Exhaustivity(X;

C)+W4×

Uniqueness(X;

C),

wherein W1, W2, W3 and W4 denote weight coefficients to indexes which are determined in advance,

wherein the Exhaustivity(X;

C) is given using a formula;

Exhaustivity(X;

C)=1/DataNum×

|U Data(ci)|,

wherein DataNum denotes an amount of data being classified,

wherein Data(ci) denotes a set of data allocated to a classification ci selected from among the plurality of classifications, and

wherein |U Data (ci)| denotes an amount of data allocated to the plurality of classifications.

28. A non-transitory computer-readable storage medium storing a data classifier program, executable by a computer, which refers to a classified hierarchy including a plurality of classifications to produce a plurality of classification axes based on classifications and data groups input, said data classifier program implementing;

a basic category accumulation process which stores the classified hierarchy establishing a parent-child relationship between the plurality of classifications and classifications serving as basic categories which are used to select desired classifications;

a multidimensional classification axis candidate creation process which creates classification axis candidates based on combinations of classifications belonging to each basic category to produce multidimensional classification axis candidates each combining a plurality of classification axis candidates; and

a multidimensional priority calculation process which inputs the multidimensional classification axis candidates to calculate priorities with respect to the multidimensional classification axis candidates with reference to hierarchical distances, each representing a length of a path reaching a common ancestor among classifications associated with the classification axis candidates, and hierarchical depths each representing a length of a path between classifications associated with the classification axis candidates,

wherein semantic independence of each of the plurality of classification axes is based on length of hierarchical distances of each of the plurality of classification axes,

wherein the multidimensional priority calculation process calculates the priority concerning a basic category X related to a plurality of classifications C in the classified hierarchy using a formula;

Priority(X;

C)=W1×

Independence(X;

C)+W2×

Specifics(X;

C)+W3×

Exhaustivity(X;

C)+W4×

Uniqueness(X;

C),

wherein W1, W2, W3 and W4 denote weight coefficients to indexes which are determined in advance,

wherein the Uniqueness(X;

C) is given using a formula;

Uniqueness(X;

C)=1/(1/|U Data(ci)|×

Σ

CatNum(ci)),

wherein Data(ci) denotes a set of data allocated to a classification ci selected from among the plurality of classifications,

wherein |U Data (ci)| denotes an amount of data having no redundancy allocated to each of the plurality of classifications, and

wherein CatNum(ci) denotes an amount of data allocated to the classification ci.