System and method for electronic communication management

• US 20040254904A1
• Filed: 05/05/2004
• Published: 12/16/2004
• Est. Priority Date: 01/03/2001
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Cohen, Dani, Hajaj, Nissan, Magdalen, Josemina, Nelken, Yoram
• Granted Patent: US 7,644,057 B2
• Time in Patent Office: Days
• US Class Current: 706/50
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 40/205 Parsing
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06Q 10/00 Administration; Management
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

A system and method for classifying text includes a pre-processor, a knowledge base, and a statistical engine. The pre-processor identifies concepts in the text and creates a structed text object that contains the concepts. structured text object is then passed to a statistical engine, which applies statistical information provided in nodes of a knowledge base to the structured text object in order to calculate a set of match scores, each match score representing the relevance of the text to an associated one of a plurality of predefined categories. The pre-processor may be implemented in the form of an interpreter which selects and executes a script that includes language- and scenario-specific instructions for performing linguistic and semantic analysis of the text.

Claims

1. A computerized text classifier system, comprising:

a pre-processor configured to analyze a text to identify concepts and generate a concept model containing the identified concepts;

a knowledge base having a plurality of nodes including a set of learning nodes, each of the learning nodes being provided with statistical information corresponding to a category; and

a statistical engine configured to calculate a set of match scores for the concept model by using the knowledge base, each match score of the set of match scores having an associated category with a suggested action and being representative of a relevance of the text to the associated category.

2. The computerized text classifier system of claim 1, wherein the text includes a plurality of fields, a first subset of the plurality of fields consisting of unstructured data and a second subset of the plurality of fields consisting of structured data.

3. The computerized text classifier system of claim 1, wherein the plurality of nodes further includes a set of rule-based nodes.

4. The computerized text classifier system of claim 1, wherein the plurality of nodes are organized into a tree structure.

5. The computerized text classifier system of claim 1, wherein the match scores are calibrated to values of an operational parameter.

6. The computerized text classifier system of claim 5, wherein the operational parameter is selected from a group consisting of precision and recall.

7. The computerized text classifier system of claim 1, wherein the pre-processor selects a script from a plurality of scripts and executes the selected script to identify concepts.

8. The computerized text classifier system of claim 7, wherein at least two of the plurality of scripts correspond to different languages.

9. The computerized text classifier system of claim 1, wherein the statistical engine is further configured to receive real-time feedback to adapt the statistical information provided to one or more learning nodes of the set of learning nodes.

10. The computerized text classifier system of claim 9, wherein the real-time feedback comprises a response of a human agent to the relevance of the text to associated categories based upon the set of match scores.

11. The computerized text classifier system of claim 9, wherein the real-time feedback comprises a reply to the suggested action, the suggested action comprising a suggested response or a link to a web-resource.

12. The computerized text classifier system of claim 9, wherein the statistical engine is further configured to modify weights associated with the statistical information, in accordance with the received real-time feedback.

13. A method of classifying text on a computer, comprising steps of:

analyzing a text to identify concepts and building a concept model containing the concepts;

providing a knowledge base having a plurality of nodes including a set of learning nodes, each of the learning nodes being provided with statistical information corresponding to a category; and

calculating a set of match scores for the concept model by using the knowledge base, each match score of the set of match scores having an associated category with a suggested action and being representative of a relevance of the text to the associated category.

14. The method of claim 13, wherein the text includes a plurality of fields, a first subset of the plurality of fields consisting of unstructured data and a second subset of the plurality of fields consisting of structured data.

15. The method of claim 13, wherein the plurality of nodes further includes a set of rule-based nodes.

16. The method of claim 13, wherein the plurality of nodes are organized into a tree structure.

17. The method of claim 13, further comprising a step of calibrating match scores to values of an operational parameter.

18. The method of claim 17, wherein the operational parameter is selected from a group consisting of precision and recall.

19. The method of claim 13, further comprising a step of selecting an appropriate script from a plurality of scripts and executing the selected script to identify concepts in the text.

20. The method of claim 19, wherein the step of selecting an appropriate script from a plurality of scripts includes identifying a language in which the text is written, and selecting the script corresponding to the identified language.

21. The method of claim 13, further comprising a step of using real-time feedback to modify the statistical information provided to one or more learning nodes of the set of learning nodes.

22. The method of claim 21, wherein the real-time feedback comprises a response of a human agent to the relevance of the text to associated categories based upon the set of match scores.

23. The method of claim 21, wherein the real-time feedback comprises a reply to the suggested action, the suggested action comprising a suggested response or a link to a web-resource.

24. The method of claim 21, wherein the step of using real-time feedback to modify the statistical information comprises a step of modifying weights assigned to the statistical information.

25. A system for classifying text on a computer, comprising:

means for analyzing a text to identify concepts and building a concept model containing the concepts;

means for providing a knowledge base having a plurality of nodes including a set of learning nodes, each of the learning nodes being provided with statistical information corresponding to a category; and

means for calculating a set of match scores for the concept model by using the knowledge base, each match score of the set of match scores having an associated category with a suggested action and being representative of a relevance of the text to the associated category.