System and method for message process and response

• US 6,278,996 B1
• Filed: 03/30/1998
• Issued: 08/21/2001
• Est. Priority Date: 03/31/1997
• 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: Aleksandrovsky, Boris, Richardson, Keith D., Greif, Jeff, Buedel, Doug
• Primary Examiner(s): Black, Thomas
• Assistant Examiner(s): RONES, CHARLES
• Application Number:
  US09/052,520
• Time in Patent Office: 1,240 Days
• Field of Search: 704/9, 704/10, 705/42, 706/25, 706/46, 709/204, 709/207, 345/339, 370/407, 395/200.31, 434/350, 128/905, 707/104, 707/5, 707/6
• US Class Current: 1/1
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/56 Natural language generation
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

A message understanding and response system recognizes and answers messages based on the message writer'"'"'s intent in unconstrained natural language text messages. The system has a set of knowledge bases with linked domain specific words, phrases, and regular expressions relating to the domain of the writer and the domain of the respondent. The writer'"'"'s domain is represented by special purpose lexicons linked to representations of typical intents. The typical intents are linked to a domain knowledge base of typical and appropriate respondent actions. The system is initialized by manually classifying a training text corpus according to the respondent'"'"'s policies. A lexical analysis tool with prototypical intents and phrases indicating intents is applied to the training text corpus, which includes the domain specific characteristics of both the writer and the respondent. The output results are an operable knowledge base which is a conjunction of keywords used to communicate between the two domains of the writer and the respondent. During automatic operation, the input text is pre-processed to remove irregularities in a manner similar to how the data in the training text corpus was regularized. Sets of extracted keywords and concepts are matched against the sets of stored, pre-classified keywords and concepts, producing a list of intents. The intents and other extracted features are then mapped to appropriate actions as defined by the system operator. The actions use the common linked domain knowledge terms to formulate a textual reply which is tailored to and answering the intent of writer of the input message.

Claims

1. A method of handling electronic mail messages based on message content, comprising the steps of:

inputting a message;

generating from content of the message a set of tokens representing the content which tokens are tagged and regularized;

matching each token in the set of tokens against a lexicon, resulting in a set of standard lexical tokens for the message;

comparing the set of standard lexical tokens to phrase expressions in a collection of lexicon entries, wherein each lexicon entry comprises a phrase expression and an associated phrase type;

identifying writer intents from the associated phrase types found in the step of comparing;

selecting, from a set of action records mapping intents to actions, action records which are mapped from at least one intent and includes an associated set of actions, based on matching message intents;

generating a self-consistent subset of the selected actions; and

executing the self-consistent subset of the selected actions.

2. The method of claim 1, wherein handling a message is a process of selecting at least one of responding to the message, routing the message or disposing of the message.

3. The method of claim 1, wherein the lexicon comprises correctly spelled words and base word forms.

4. The method of claim 1, wherein the phrase type is selected from a word, a plurality of words, a word phrase, a plurality of word phrases, a regular expression text pattern, a plurality of regular expression text patterns, a message formatting, a plurality of message formatting or a set of phrase types.

5. The method of claim 1, wherein the self-consistent subset of the selected actions comprises all of the selected actions when all of the selected actions are self-consistent.

6. The method of claim 1, wherein the at least one intent from which an action record is mapped is either a writer intent or a responder intent.

7. The method of claim 6, wherein the responder intent is derived from business rules of an operator of a system handling the message.

8. The method of claim 1, wherein the at least one message classification reflects observed conditions or deduced conditions independent of the writer intents.

9. The method of claim 1, further comprising a step of evaluating message content to identify missing phrase types which are candidates to be added to the lexicon.

10. The method of claim 1, further comprising the steps of:

identifying at least one message classification from the content of the message;

selecting additional actions based on the at least one message classification; and

including the additional actions in the step of generating a self-consistent subset of actions.

11. The method of claim 1, wherein each action record represents a response to a writer intent and is selected from a direct response to the writer, a routing referral to a human reader, or a call to an external computer program.

12. The method of claim 1, further comprising the steps of:

separating free form text in the content from structured text in the content, prior to the step of generating the set of tokens; and

detecting free form text embedded in the structured text, prior to the step of generating the set of tokens, wherein the set of tokens included tokens representing the embedded free form text.

13. The method of claim 1, wherein the step of identifying writer intents comprises the steps of:

matching identified tokens to a database of regular expression patterns; and

matching identified tokens to a database of phrase types other than regular expression patterns.

14. The method of claim 13, wherein the identified tokens comprise regularized tokens and tokens prior to regularization.

15. The method of claim 1, further comprising the steps of:

matching the identified phrase types and message formatting types to the phrase types and message formatting types of prototype message intents described by groups of associated phrase and message formatting types.