System, method and computer program product for data mining in a normalized relational database

• US 6,691,120 B1
• Filed: 06/30/2000
• Issued: 02/10/2004
• Est. Priority Date: 06/30/2000
• Status: Expired due to Term

• Current Assignee: Teradata US, Inc. (Teradata Corporation)
• Original Assignee: NCR Corporation (NCR Voyix Corporation)
• Inventors: Aldridge, Bruce E., Durrant, Douglas J.
• Primary Examiner(s): Metjahic, Safet
• Assistant Examiner(s): Nguyen, Cindy
• Application Number:
  US09/608,085
• Time in Patent Office: 1,320 Days
• Field of Search: 707/100, 707/2, 707/4, 707/104.1
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06F 16/283 Multi-dimensional databases or data warehouses, e.g. MOLAP or ROLAP
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A method and system for preparing a data record with multidimensional attributes for processing by a data-mining tool includes the steps of identifying a normalized data record, which includes an existing attribute comprising a plurality of dimensions, and generating a plurality of new attributes corresponding to the plurality of dimensions of the existing attribute. The method further includes creating an REI data record comprising the plurality of new attributes. The method further includes associating a plurality of value instances corresponding to the dimensions of the existing attribute with the plurality of new attributes.

Claims

1. In a data-mining system for use with a database stored in a normalized form, a method for preparing a data record with multidimensional attributes for processing by a data-mining tool, the method comprising the steps of

identifying a normalized data record, which includes an existing attribute comprising a plurality of dimensions;

generating a plurality of new attributes corresponding to the plurality of dimensions of the existing attribute;

creating an REI data record comprising the plurality of new attributes; and

associating a plurality of value instances corresponding to the dimensions of the existing attribute with the plurality of new attributes.

2. The method of claim 1, wherein said identifying step further comprises the step of examining a constraint for an attribute.

3. The method of claim 2, wherein said examining step further comprises examining a domain for the attribute.

4. The method of claim 1, wherein said generating step further comprises the step of making names for the new attributes, the names being a composite of an existing name for the existing attribute and a set of domain values for a dimension attribute corresponding to the plurality of dimensions.

5. The method of claim 1, wherein the REI data record further comprises additional attributes and said creating step further comprises creation of an REI table comprising the plurality of new attributes and the additional attributes.

6. The method of claim 1, wherein said associating step further comprises the step of determining indices that refer to the normalized data record.

7. In a data-mining system for use with a database stored in a normalized form, a method for transforming a subset of the database, with multidimensional attributes, into an REI table for processing by a data-mining tool, the method comprising the steps of

selecting a data-item key;

identifying a composite key comprising the data-item key and a local key attribute;

determining a dimensionality of the local key attribute;

generating new non-key attributes based upon the dimensionality of the local key attribute and non-composite-key attributes;

creating an REI table defined by the data-item key and the new non-key attributes; and

associating value instances with the REI table.

8. The method of claim 7, wherein said selecting step further comprises the step of querying a user.

9. The method of claim 7, wherein said selecting step further comprises the step of finding a pervasive attribute that is part of all tables making up a database subset being transformed.

10. The method of claim 7, wherein said identifying step further comprises the step of examining a constraint for the local key attribute.

11. The method of claim 7, wherein said identifying step further comprises the step of obtaining input from a user.

12. The method of claim 7, wherein said determining step further comprises the step of examining a constraint for the local key attribute.

13. The method of claim 12, wherein said examining step further comprises examining a domain for the local key attribute.

14. The method of claim 7, wherein said generating step further comprises the step of making names for the new non-key attributes, the names being a combination of non-composite-key names for the non-composite-key attributes and a set of domain values for the local key attribute.

15. The method of claim 7, wherein the REI table is further defined by additional attributes.

16. The method of claim 7, wherein said associating step further comprises the step of saving indices that refer to the value instances.

17. The method of claim 7, wherein said associating step further comprises the steps of examining a special-case non-key attribute;

and

excluding from said associating step all data rows having a non-priority value for the special-case non-key attribute.

18. A computer program product for use in a data-mining system, said computer program product including a medium usable by a processor, the medium having stored thereon a sequence of instructions which, when executed by the processor, causes the processor to execute a process that pivots multidimensional attributes thereby enabling the data-mining system to analyze data stored in a normalized database having multidimensional attributes, said computer program product comprising:

instructions for identifying a composite key comprising a data-item key and a local key attribute;

instructions for generating new non-key attributes based upon a dimensionality of the local key attribute and non-composite-key attributes;

instructions for creating an REI table defined by the data-item key and the new non-key attributes; and

instructions for associating value instances with the REI table.

19. The computer program product of claim 18, wherein said instructions for identifying further comprise:

instructions for querying a user to select the data-item key;

instructions for enabling the user to identify the local key attribute;

instructions for finding a pervasive attribute to be used as the data-item key; and

instructions for examining constraint and domain information for the local key attribute.

20. The computer program product of claim 18, wherein said instructions for generating

further comprise instructions for making names for the new non-key attributes, the names being a combination of non-composite-key names for the non-composite-key attributes and a set of domain values for the local key attribute.

21. The computer program product of claim 18, wherein the REI table is further defined by additional attributes.

22. The computer product of claim 18, wherein said instructions for associating further comprise instructions for saving indices that refer to the value instances.

23. The computer program product of claim 18, wherein said instructions for associating further comprise:

instructions for examining a special case non-key attribute; and

instructions for excluding unwanted value instances from the REI table, when the unwanted value instances belong to a data row having a non-priority value for the special-case non-key attribute.