Automatic electronic message interpretation and routing system

• US 6,182,059 B1
• Filed: 05/08/1997
• Issued: 01/30/2001
• Est. Priority Date: 04/03/1997
• Status: Expired due to Term

• Current Assignee: Bright Response, LLC (f/k/a Polaris IP LLC) (Legacy Patent Holdings, LLC)
• Original Assignee: Brightware Incorporated (Oracle Corporation)
• Inventors: Hsu, Julie, Rice, Amy, Angotti, Anthony A., Piccolo, Rosanna M., Cohen, Fred R.
• Primary Examiner(s): Hafiz, Tariq R.
• Assistant Examiner(s): Khatri, Anil
• Application Number:
  US08/853,074
• Time in Patent Office: 1,363 Days
• Field of Search: 706/45, 706/46, 706/47, 706/49
• US Class Current: 706/45
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]

Abstract

A method for automatically interpreting an electronic message, including the steps of (a) receiving the electronic message from a source; (b) interpreting the electronic message using a rule base and case base knowledge engine; and (c) classifying the electronic message as at least one of (i) being able to be responded to automatically; and (ii) requiring assistance from a human operator. The method for automatically interpreting an electronic message may also include the step of retrieving one or more predetermined responses corresponding to the interpretation of the electronic message from a repository for automatic delivery to the source.

Claims

1. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(c) classifying the electronic message as at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator;

(c1) further categorizing the electronic message into at least one of a plurality of sub-categories based on subject matter content of the electronic message;

(c2) prioritizing the sub-categorized electronic message into at least one of a plurality of priorities based on the subject matter content of the electronic message wherein a higher priority indicates that the human operator should process the associated electronic message before processing lower prioritized electronic messages;

(d) retrieving one or more predetermined responses from a repository, the predetermined responses being proposed for delivery to the source;

(e) forwarding the electronic message and the predetermined response to the human operator when the classification step indicates that a response to the electronic message requires assistance from a human operator; and

(f) delivering the predetermined response to the source when the human operator deems the response appropriate,

wherein the plurality of priorities of a product service sub-category include at least one of (i) fraud and lost products;

(ii) sensitive information;

(iii) general information; and

(iv) user comments.

2. The method of claim 1, wherein the listed priorities are in order from highest to lowest priority.

3. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(c) classifying the electronic message as at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator;

(c1) further categorizing the electronic message into at least one of a plurality of sub-categories based on subject matter content of the electronic message;

(c2) prioritizing the sub-categorized electronic message into at least one of a plurality of priorities based on the subject matter content of the electronic message wherein a higher priority indicates that the human operator should process the associated electronic message before processing lower prioritized electronic messages;

(d) retrieving one or more predetermined responses from a repository, the predetermined responses being proposed for delivery to the source;

(e) forwarding the electronic message and the predetermined response to the human operator when the classification step indicates that a response to the electronic message requires assistance from a human operator; and

(f) delivering the predetermined response to the source when the human operator deems the response appropriate,

wherein the plurality of priorities of a product sales sub-category include promotional content, request for services, and general questions and lengthy messages.

4. The method of claim 3, wherein the listed priorities are in order from highest to lowest priority.

5. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine; and

(c) classifying the electronic message as at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator, the assistance including at least a first level of assistance and a second level of assistance;

when the classification step indicates that the electronic message requires a second level of assistance from a human operator, the method further comprising the steps of;

(d) retrieving one or more predetermined remarks from a remarks repository to assist the human operator in processing the electronic message manually; and

(e) forwarding the electronic message to the human operator,

wherein the classification step indicates that the electronic message requires a second level of assistance from a human operator when at least one of a phone number, a foreign address, a do not call request, a facsimile number, a specific employee request, sensitive information, and a specific manual procedure is interpreted in the electronic message.

6. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(b1) classifying the electronic message as at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator;

(b2) producing a case model of the electronic message including a set of predetermined attributes for identifying specific features of the electronic message;

(b3) detecting at least one of text, combinations of text, and patterns of text of the electronic message using character matching;

(b4) flagging the attributes of the case model which are detected in the electronic message;

(b5) classifying the electronic message as at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator, the classification being performed in accordance with the flagged attributes; and

(c) retrieving one or more predetermined responses corresponding to the interpretation of the electronic message from a repository for automatic delivery to the source when the classification step indicates that the electronic message can be responded to automatically.

7. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(b1) classifying the electronic message as at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator;

(b2) producing a case model of the electronic message including (i) a set of attributes for identifying specific features of the electronic message; and

(ii) message text;

(b3) detecting at least one of text, combinations of text, and patterns of text of the electronic message using character matching;

(b4) flagging the attributes of the case model which are detected in the electronic message;

(b5) comparing the flagged attributes of the case model with stored attributes of stored case models of the case base;

(b6) comparing the text of the case model with stored text of the stored case models of the case base;

(b7) assigning a score to each stored case model which is compared with the case model, the score increasing when at least one of the attributes and the text match the stored case model and the score not increasing when at least one of the attributes and the text do not match the stored case model; and

(c) retrieving one or more predetermined responses corresponding to the interpretation of the electronic message from a repository for automatic delivery to the source when the classification step indicates that the electronic message can be responded to automatically.

8. The method of claim 7, wherein:

when at least one of the attributes and the text match the stored case model, the score is increased by a predetermined match weight; and

when at least one of the attributes and the text does not match the stored case model, the score is decreased by a predetermined mismatch weight.

9. The method of claim 8, wherein the match weight has an absolute value greater than zero and the mismatch weight is zero.

10. The method of claim 8, wherein each score is normalized by dividing the score by a maximum possible score for the stored case model, where the maximum possible score is determined when all of the attributes and text of the case model and the stored case model match.

11. A system for automatically interpreting a non-interactive electronic message received from a source, the system comprising:

a server for transmitting and receiving electronic messages over a communications channel;

an inbox storage device for storing incoming electronic messages;

a knowledge engine including a rule base and a case base, the case base having a plurality of stored cases representing past received electronic messages;

a pre-processor for receiving the electronic message and interpreting the electronic message using the rule base;

a searching device for searching the electronic message and the case base to retrieve a stored case from the case base which most closely matches the electronic message;

a classifier for classifying the electronic message into at least one of (i) being able to be responded to automatically; and

(ii) requiring assistance from a human operator;

a repository of predetermined responses, one or more of the predetermined responses being selected by the knowledge base for proposed delivery to the source; and

an electronic router for forwarding the electronic message to the human operator when the classifier indicates that a response to the electronic message requires assistance from a human operator, the router delivering the predetermined response to the source when the human operator deems the response appropriate.

12. The system of claim 11, wherein the classifier categorizes the electronic message into at least one of a plurality of sub-categories based on subject matter content of the electronic message;

wherein the classifier prioritizes the sub-categorized electronic message into at least one of a plurality of priorities based on the subject matter content of the electronic message wherein a higher priority indicates that the human operator should process the associated electronic message before processing lower prioritized electronic messages; and

wherein the plurality of priorities of a product service sub-category include at least one of (i) fraud and lost products;

(ii) sensitive information;

(iii) general information; and

(iv) user comments.

13. The system of claim 12, wherein the listed priorities are in order from highest to lowest priority.

14. The system of claim 11, wherein the classifier categorizes the electronic message into at least one of a plurality of sub-categories based on subject matter content of the electronic message;

wherein the classifier prioritizes the sub-categorized electronic message into at least one of a plurality of priorities based on the subject matter content of the electronic message wherein a higher priority indicates that the human operator should process the associated electronic message before processing lower prioritized electronic messages; and

wherein the plurality of priorities of a product sales sub-category include promotional content, request for services, and general questions and lengthy messages.

15. The system of claim 14, wherein the listed priorities are in order from highest to lowest priority.

16. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(b1) categorizing the electronic message into at least one of a plurality of sub-categories based on subject matter content of the electronic message, wherein the sub-categories include product service subject matter and product sales subject matter;

(b2) prioritizing the sub-categorized electronic message into at least one of a plurality of priorities based on the subject matter content of the electronic message wherein a higher priority indicates that the human operator should process the associated electronic message before processing lower prioritized electronic messages;

(c) retrieving one or more predetermined responses from a repository, the predetermined responses being proposed for delivery to the source;

(d) forwarding the electronic message and the predetermined response to a human operator; and

(e) delivering the predetermined response to the source when the human operator deems the response appropriate,

wherein the plurality of priorities of a product service sub-category include at least one of (i) fraud and lost products;

(ii) sensitive information;

(iii) general information; and

(iv) user comments.

17. The method of claim 16, wherein the listed priorities are in order from highest to lowest priority.

18. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(b1) categorizing the electronic message into at least one of a plurality of sub-categories based on subject matter content of the electronic message, wherein the sub-categories include product service subject matter and product sales subject matter;

(b2) prioritizing the sub-categorized electronic message into at least one of a plurality of priorities based on the subject matter content of the electronic message wherein a higher priority indicates that the human operator should process the associated electronic message before processing lower prioritized electronic messages;

(c) retrieving one or more predetermined responses from a repository, the predetermined responses being proposed for delivery to the source;

(d) forwarding the electronic message and the predetermined response to a human operator; and

(e) delivering the predetermined response to the source when the human operator deems the response appropriate,

wherein the plurality of priorities of a product sales sub-category include promotional content, request for services, and general questions and lengthy messages.

19. The method of claim 18, wherein the listed priorities are in order from highest to lowest priority.

20. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(b1) producing a case model of the electronic message including a set of predetermined attributes for identifying specific features of the electronic message;

(b2) detecting at least one of text, combinations of text, and patterns of text of the electronic message using character matching;

(b3) flagging the attributes of the case model which are detected in the electronic message;

(c) retrieving one or more predetermined responses from a repository, the predetermined responses being proposed for delivery to the source;

(d) forwarding the electronic message and the predetermined response to a human operator; and

(e) delivering the predetermined response to the source when the human operator deems the response appropriate.

21. A method for automatically interpreting a non-interactive electronic message, comprising the steps of:

(a) receiving the electronic message from a source;

(b) interpreting the electronic message using a rule base and case base knowledge engine;

(b1) producing a case model of the electronic message including (i) a set of attributes for identifying specific features of the electronic message; and

(ii) message text;

(b2) detecting at least one of text, combinations of text, and patterns of text of the electronic message using character matching;

(b3) flagging the attributes of the case model which are detected in the electronic message;

(b4) comparing the flagged attributes of the case model with stored attributes of stored case models of the case base;

(b5) comparing the text of the case model with stored text of the stored case models of the case base;

(b6) assigning a score to each stored case model which is compared with the case model, the score increasing when at least one of the attributes and the text match the stored case model and the score not increasing when at least one of the attributes and the text do not match the stored case model;

(c) retrieving one or more predetermined responses from a repository, the predetermined responses being proposed for delivery to the source;

(d) forwarding the electronic message and the predetermined response to a human operator; and

(e) delivering the predetermined response to the source when the human operator deems the response appropriate.

22. The method of claim 21, wherein:

when at least one of the attributes and the text match the stored case model, the score is increased by a predetermined match weight; and

when at least one of the attributes and the text does not match the stored case model, the score is decreased by a predetermined mismatch weight.

23. The method of claim 22, wherein each score is normalized by dividing the score by a maximum possible score for the stored case model, where the maximum possible score is determined when all of the attributes and text of the case model and the stored case model match.

24. The method of claim 22, wherein the match weight has an absolute value greater than zero and the mismatch weight is zero.