Integrating campaign management and data mining

• US 6,240,411 B1
• Filed: 06/15/1998
• Issued: 05/29/2001
• Est. Priority Date: 06/15/1998
• Status: Expired due to Term

• Current Assignee: Amdocs Development Limited (Amdocs Ltd.), Amdocs Software Systems Limited (Amdocs Ltd.), Exchange Applications, Inc.
• Original Assignee: Exchange Applications, Inc.
• Inventors: Thearling, Kurt
• Primary Examiner(s): AMSBURY, WAYNE P
• Application Number:
  US09/097,875
• Time in Patent Office: 1,079 Days
• Field of Search: 707/5, 707/6, 706/12
• 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 16/285 Clustering or classification
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

Method and apparatus are disclosed for integration of campaign management and data mining. The method and apparatus disclose incorporating references to data mining models into the campaign management process. In some embodiments, this permits evaluating the data mining model for fewer than all of the records in a database, potentially saving computation time. The method and apparatus can include building queries for a database or ranking criteria for records in a database that include a reference to a data mining model.

Claims

1. A method of classifying a plurality records in a database, the method comprising steps of:

(a) providing a first model for ascertaining a first characteristic of individual ones of the records;

(b) forming a selection criteria, the selection criteria including a reference to the first model, wherein the step (b) comprises a step of forming a query, the query including the reference to the first model;

(c) using the reference to execute the first model to generate a value for the first characteristic at least one of the plurality of records; and

(d) selecting a selected set of the records, each record of the selected set satisfying the selection criteria.

2. A method of classifying a plurality records in a database, the method comprising steps of:

(a) providing a first model for ascertaining a first characteristic of individual ones of the records;

(b) forming a selection criteria, the selection criteria including a reference to the first model;

(c) using the reference to execute the first model to generate a value for the first characteristic at least one of the plurality of records; and

(d) selecting a selected set of the records, each record of the selected set satisfying the selection criteria, wherein the step (d) comprises a step of ranking the records.

3. A method of classifying a plurality records in a database, the method comprising steps of:

(a) providing a first model for ascertaining a first characteristic of individual ones of the records;

(b) forming a selection criteria, the selection criteria including a reference to the first model;

(c) using the reference to execute the first model to generate a value for the first characteristic at least one of the plurality of records; and

(d) selecting a selected set of the records, each record of the selected set satisfying the selection criteria;

wherein the records correspond to entities and the method further comprises steps of;

gathering entity data for a plurality of entities to include in the database, each record in the database corresponding to one of the entities; and

using the selected set of records to determine for which of the entities to perform a corresponding action.

4. A method of classifying a plurality records in a database, the method comprising steps of:

(a) providing a first model for ascertaining a first characteristic of individual ones of the records;

(b) forming a selection criteria, the selection criteria including a reference to the first model, and wherein the step (b) comprises a step of

forming a query to select the records, the query including the reference to the first model and a reference to the second model;

(c) using the reference to execute the first model to generate a value for the first characteristic at least one of the plurality of records; and

(d) selecting a selected set of the records, each record of the selected set satisfying the selection criteria;

further comprising a step of;

providing a second model for ascertaining a second characteristic of individual ones of the records.

5. A method of classifying a plurality records in a database, the method comprising steps of:

(a) providing a first model for ascertaining a first characteristic of individual ones of the records;

(b) forming a selection criteria, the selection criteria including a reference to the first model;

(c) using the reference to execute the first model to generate a value for the first characteristic at least one of the plurality of records, wherein;

the step (c) comprises a step of

(e) evaluating the first model for fewer than all of the records in the database; and

(d) selecting a selected set of the records, each record of the selected set satisfying the selection criteria, wherein the step (d) comprises a step of

(f) resolving the selection criteria without evaluating the first model for all of the records in the database.

6. The method of claim 5, wherein the step (e) comprises steps of:

(g) building a model evaluation database, the model evaluation database including a subset of the records of the database; and

evaluating the first model only for the records in the model evaluation database.

7. The method of claim 6, wherein the selection criteria is a query and the step (g) comprises a step of evaluating a portion of a query.

8. A campaign management system to classify records in a database, the system comprising:

a selection criteria processor to process a selection criteria, wherein resolution of the selection criteria requires evaluation of a model;

a model evaluator, responsive to the selection criteria processor, to evaluate the model; and

means for building a campaign; and

wherein

the means for building comprising means for formulating the selection criteria.

9. The system of claim 8, further comprising:

a model library to provide the model.

10. A campaign management system to classify records in a database, the system comprising:

a selection criteria processor to process a selection criteria, wherein resolution of the selection criteria requires evaluation of a model; and

a model evaluator, responsive to the selection criteria processor, to evaluate the model;

wherein;

the model evaluator comprises means for evaluating the model for fewer than all of the records in the database; and

the selection criteria processor comprises means for resolving a query including a reference to the model, without evaluating the model for all of the records in the database.

11. The system of claim 10, wherein the means for resolving comprises:

means for building a model evaluation database, the model evaluation database including a subset of the records of the database and having fewer records than the database; and

means for evaluating the model only for the records in the model evaluation database.

12. A campaign management system to classify records in a database, the system comprising:

a selection criteria processor to process a selection criteria, wherein resolution of the selection criteria requires evaluation of a model;

a model evaluator, responsive to the selection criteria processor, to evaluate the model; and

a model library to provide the model.

13. A method of selecting records in a computer database storing a plurality of records, the method comprising steps of:

(a) providing a model for ascertaining a characteristic of individual ones of the records;

(b) forming a query to evaluate the records, the query including a reference to the model, wherein the query includes a plurality of elements, one of the elements including the reference to the model;

(c) selecting a selected set of records from the plurality of records, each record of the selected set satisfying the query, wherein the step (c) comprises a step of (e) resolving at least one of the elements before the step (d); and

(d) during the step (c), evaluating the model for at least one of the plurality of records, wherein the step (d) comprises steps of

forming a restricted group of the records based on the step (e), the restricted group including less than all of the records in the database, and

evaluating the model only for the records in the restricted group.

14. The method of claim 13, wherein:

the at least one element resolved in step (e) is in the query as part of an AND clause including the model reference; and

the step (c) further comprises steps of

removing records from the restricted group that do not satisfy the element including the reference to the model, and

using the restricted group with removed records as a partial result for resolution of the AND clause.

15. A campaign management system to classify a plurality of records in a database, the system comprising:

means for receiving a selection criteria that includes a model reference;

means, responsive to the selection criteria, for selecting a selected set of records from the plurality of records, each record of the selected set of records satisfying the query, wherein the means for selecting comprises means for resolving at least one of a plurality of elements in a query, before resolving the model reference; and

means, responsive to the selection criteria, for resolving the model reference to produce a model value for at least one of the plurality of records, wherein the means for resolving the model reference comprises means, responsive to the means for resolving at least one of a plurality of elements, for forming a restricted group of the records, and means for determining the model values only for the records in the restricted group.

16. A method of resolving a selection criteria for a database storing a plurality of records, the selection criteria including a reference to a model for ascertaining a characteristic of individual ones of the records, the method comprising steps of:

(a) evaluating the model for a restricted set of the records in the database;

(b) resolving the selection criteria without evaluating the model for all of the records in the database, to produce a selected set of the records; and

(c) saving the result of evaluating the model.

17. A method of resolving a selection criteria for a database storing a plurality of records, the selection criteria including a reference to a model for ascertaining a characteristic of individual ones of the records, the method comprising steps of:

(a) evaluating the model for a restricted set of the records in the database wherein the records correspond to entities;

(b) resolving the selection criteria without evaluating the model for all of the records in the database, to produce a selected set of the records; and

(c) gathering entity data for a plurality of entities to include in the database, each record in the database corresponding to one of the entities; and

(d) using the selected set of records to determine for which of the entities to perform a corresponding action.

18. A method of resolving a selection criteria for a database storing a plurality of records, the selection criteria including a reference to a model for ascertaining a characteristic of individual ones of the records, the method comprising steps of:

(a) evaluating the model for a restricted set of the records in the database, wherein the step (a) comprises steps of;

(c) building a model evaluation database, the model evaluation database including a subset of the records of the database; and

(d) evaluating the model only for the records in the model evaluation database; and

(b) resolving the selection criteria without evaluating the model for all of the records in the database, to produce a selected set of the records, wherein the selection criteria is a query.

19. The method of claim 18, wherein the records correspond to entities and the method further comprises steps of:

gathering entity data for a plurality of entities to include in the database, each record in the database corresponding to one of the entities; and

using the selected set of records to determine for which of the entities to perform an action.

20. The method of claim 18, wherein:

the query includes a first element test, the first element test being related to the reference through an AND relationship in the query; and

the step (c) comprises a step of

identifying the records of the database that meet the first element test.

21. The method of claim 20, wherein:

the query includes a second element test, the second element test being related to the reference through an OR relationship in the query; and

the step (c) comprises a step of

identifying the records of the database that do not meet the second element test.

22. The method of claim 18, wherein:

the query includes an element test, the element test being related to the reference through an OR relationship in the query; and

the step (c) comprises a step of

identifying the records of the database that do not meet the element test.

23. A method of forming a selection criteria for a database including a plurality of records, the method comprising steps of:

providing a library including at least one data mining model used to score the records in the database;

selecting at least one data mining model for use in the selection criteria; and

constructing the selection criteria using a reference to the selected data mining model, wherein the selection criteria is a query.

24. A method of forming a selection criteria for a database including a plurality of records, the method comprising steps of:

providing a library including at least one data mining model used to score the records in the database;

selecting at least one data mining model for use in the selection criteria; and

constructing the selection criteria using a reference to the selected data mining model, wherein the reference is stored in a physical medium in a format selected from the group consisting of;

program code for the model, a pointer to another location in a second physical medium, corresponding to program code for the model, and a set of parameters for evaluation of the model.

27. The method of claim 24, wherein the reference is stored in a physical medium in a format selected from the group consisting of:

program code for the model, a pointer to another location in a second physical medium, corresponding to program code for the model, and a set of parameters for evaluation of the model.

25. A method of using a campaign management system, comprising steps of:

(a) initiating the campaign management system;

(b) using the campaign management system to access a model, wherein the step (b) comprises a step of providing a reference to the model; and

(c) scoring records in a database using the model.

26. A method of using a campaign management system, comprising steps of:

(a) initiating the campaign management system;

(b) using the campaign management system to access a model;

(c) scoring records in a database using the model; and

(d) providing a model library, the model library including a reference to the model.