Method and apparatus providing computer understanding and instructions from natural language

• US 20020169597A1
• Filed: 01/11/2002
• Published: 11/14/2002
• Est. Priority Date: 03/12/2001
• Status: Abandoned Application

• Current Assignee: Fain Systems, Inc.
• Original Assignee: Fain Systems, Inc.
• Inventors: Fain, Samuel V., Fain, Vitaliy S.
• Application Number:
  US10/043,998
• Publication Number:
  US 20020169597A1
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

Computer understanding and generation of computer instructions from natural language dialog utilizes using processes and data structures that map natural language utterances to computer program modules. A series of dictionaries, including a subject area dictionary, a program module subdictionary, an argument subdictionary and a value subdictionary are built and used by a computer instruction generator program to map natural language utterances to computer instructions. Selection of appropriate computer program modules is performed using matching algorithms and is enhanced using historically successful probability-based data.

Claims

1. A method for providing computer understanding by generating computer instructions from a natural language dialog, comprising:

receiving a symbolic representation of a natural language utterance;

determining, by accessing a context sensitive system dictionary for a subject area, a subject area identifier based upon parsing the symbolic representation, the parsing producing parsed information;

determining, by accessing a context sensitive system subdictionary for a program module of the subject area, a module identifier based upon the determined subject area identifier and the parsed information;

determining, by accessing a context sensitive system subdictionary for an argument of the program module, an argument identifier based upon the determined module identifier and the parsed information;

determining, by accessing a context sensitive system subdictionary for a value of the argument, a value identifier based upon the determined argument identifier and the parsed information; and

producing computer instructions based upon the subject area identifier, module, the module identifier, the argument identifier and the value identifier, such that the natural language utterance is processed by the computer.

2. The method of claim 1 wherein the context sensitive system dictionary for the subject area further comprises a context sensitive system subdictionary for a sub-subject area.

3. The method of claim 1 wherein

determining a value identifier further comprises querying the computer system for a missing value identifier.

4. The method of claim 1 wherein

determining a subject area identifier further comprises querying a user of the computer system for a missing subject area identifier;

determining a module identifier further comprises querying a user of the computer system for a missing module identifier; and

determining a value identifier further comprises querying a user of the computer system for a missing value identifier.

5. The method of claim 1 wherein

determining a subject area identifier further comprises using a previously determined value for a missing subject area identifier;

determining a module identifier further comprises using a previously determined value for a missing module identifier; and

determining a value identifier further comprises using a previously determined value for a missing value identifier.

6. A method for determining an appropriate program module selection for processing a natural language dialog in a computer system for processing natural language, comprising:

capturing a set of successfully understood natural language dialogs and associated program modules used to produce computer understanding;

analyzing the captured program module information to determine a frequency of occurrence value for proceeding to a next program module from a current program module;

storing the frequency of occurrence values in a matrix; and

determining, using the matrix, the appropriate program module selection based on choosing program modules having non-zero frequency value entries in the matrix.

7. The method for claim 6 further comprising:

capturing a step associated with the program modules as executed within the natural language dialogs;

analyzing the captured program module information to determine a frequency of occurrence value, for each of the steps in the dialog, for proceeding to a next program module from a current program module;

storing the frequency of occurrence values and step information in a matrix; and

determining, using the matrix, the appropriate program module selection based on choosing program modules with matching step information and having non-zero frequency value entries in the matrix.

8. The method for claim 6 further comprising:

capturing grouping information for the program modules as executed within the natural language dialogs;

analyzing the captured program module information to determine a frequency of occurrence value, for each of the groupings, for proceeding to a next program module from a current program module;

storing the frequency of occurrence values and the grouping information in a matrix; and

determining, using the matrix, the appropriate program module selection based on choosing program module groupings having non-zero frequency value entries in the matrix.

9. An apparatus providing computer understanding by generating computer instructions from a natural language dialog, comprising:

a receiver receiving a symbolic representation of a natural language utterance;

a context sensitive subject area system dictionary used to determine a subject area identifier based upon parsing the symbolic representation, the parsing producing parsed information;

a context sensitive system program module subdictionary used to determine a module identifier based upon the determined subject area identifier and the parsed information;

a context sensitive argument system subdictionary used to determine an argument identifier based upon the determined module identifier and the parsed information;

a context sensitive value system subdictionary used to determine a value identifier based upon the determined argument identifier and the parsed information; and

computer instructions produced based upon the subject area identifier, module, the module identifier, the argument identifier and the value identifier, such that the natural language utterance is processed by the computer.

10. The apparatus of claim 9 wherein the context sensitive system dictionary for the subject area further comprises a context sensitive system subdictionary for a sub-subject area.

11. The apparatus of claim 9 wherein undetermined value identifiers are determined by querying the computer system for a missing value identifier.

12. The apparatus of claim 9 wherein:

undetermined subject area identifiers are determined by querying a user of the computer system for a missing subject area identifier;

undetermined module identifiers are determined by querying a user of the computer system for a missing module identifier; and

undetermined value identifiers are determined by querying a user of the computer system for a missing value identifier.

13. The apparatus of claim 9 wherein

undetermined subject area identifiers are determined using a previously determined value for a missing subject area identifier;

undetermined module identifiers are determined using a previously determined value for a missing module identifier; and

undetermined value identifiers are determined using a previously determined value for a missing value identifier.

15. The apparatus of claim 14 further comprising:

a step identifier, associated with the program modules as executed within the natural language dialogs;

an analyzer analyzing the captured program module information to determine a frequency of occurrence value, for each of the steps identified in the dialog, for proceeding to a next program module from a current program module;

a matrix storing the frequency of occurrence values and step information; and

a logic unit determining, using the matrix, the appropriate program module selection based on choosing program modules with matching step information and having non-zero frequency value entries in the matrix.

16. The apparatus of claim 14 further comprising:

a grouping identifier for the program modules as executed within the natural language dialogs;

an analyzer analyzing the captured program module information to determine a frequency of occurrence value, for each of the groupings, for proceeding to a next program module from a current program module;

a matrix storing the frequency of occurrence values and the grouping information; and

a logic unit determining, using the matrix, the appropriate program module selection based on choosing program module groupings having non-zero frequency value entries in the matrix.

14. An apparatus determining an appropriate program module selection for processing a natural language dialog in a computer system for processing natural language, comprising:

a set of successfully understood natural language dialogs and associated program modules used to produce computer understanding;

an analyzer analyzing the captured program module information to determine a frequency of occurrence value for proceeding to a next program module from a current program module;

a matrix storing the frequency of occurrence values; and

a logic unit determining, using the matrix, the appropriate program module selection based on choosing program modules having non-zero frequency value entries in the matrix.

17. An apparatus for providing computer understanding by generating computer instructions from a natural language dialog, comprising:

a means for receiving a symbolic representation of a natural language utterance;

a means for determining, by accessing a context sensitive system dictionary for a subject area, a subject area identifier based upon parsing the symbolic representation, the parsing producing parsed information;

a means for determining, by accessing a context sensitive system subdictionary for a program module of the subject area, a module identifier based upon the determined subject area identifier and the parsed information;

a means for determining, by accessing a context sensitive system subdictionary for an argument of the program module, an argument identifier based upon the determined module identifier and the parsed information;

a means for determining, by accessing a context sensitive system subdictionary for a value of the argument, a value identifier based upon the determined argument identifier and the parsed information; and

a means for producing computer instructions based upon the subject area identifier, module, the module identifier, the argument identifier and the value identifier, such that the natural language utterance is processed by the computer.

18. A computer program product comprising:

a computer usable medium for providing computer understanding by generating computer instructions from a natural language dialog;

a set of computer program instructions embodied on the computer usable medium, including instructions to;

receive a symbolic representation of a natural language utterance;

determine, by accessing a context sensitive system dictionary for a subject area, a subject area identifier based upon parsing the symbolic representation, the parsing producing parsed information;

determine, by accessing a context sensitive system subdictionary for a program module of the subject area, a module identifier based upon the determined subject area identifier and the parsed information;

determine, by accessing a context sensitive system subdictionary for an argument of the program module, an argument identifier based upon the determined module identifier and the parsed information;

determine, by accessing a context sensitive system subdictionary for a value of the argument, a value identifier based upon the determined argument identifier and the parsed information; and

produce computer instructions based upon the subject area identifier, module, the module identifier, the argument identifier and the value identifier, such that the natural language utterance is processed by the computer.

19. A computer data signal embodied in a carrier wave comprising a code segment for providing computer understanding by generating computer instructions from a natural language dialog, the code segment including instructions to:

receive a symbolic representation of a natural language utterance;

determine, by accessing a context sensitive system dictionary for a subject area, a subject area identifier based upon parsing the symbolic representation, the parsing producing parsed information;

determine, by accessing a context sensitive system subdictionary for a program module of the subject area, a module identifier based upon the determined subject area identifier and the parsed information;

determine, by accessing a context sensitive system subdictionary for an argument of the program module, an argument identifier based upon the determined module identifier and the parsed information;

determine, by accessing a context sensitive system subdictionary for a value of the argument, a value identifier based upon the determined argument identifier and the parsed information; and

produce computer instructions based upon the subject area identifier, module, the module identifier, the argument identifier and the value identifier, such that the natural language utterance is processed by the computer.