Expert system with natural-language rule updating

• US 4,920,499 A
• Filed: 09/30/1987
• Issued: 04/24/1990
• Est. Priority Date: 09/30/1987
• Status: Expired due to Term

• Current Assignee: Pavilion Technologies Incorporated (Rockwell Automation, Inc.)
• Original Assignee: E. I. du Pont de Nemours and Company (Corteva, Inc.)
• Inventors: Skeirik, Richard D.
• Primary Examiner(s): RUGGIERO, JOSEPH F
• Application Number:
  US07/103,050
• Time in Patent Office: 937 Days
• Field of Search: 364/513, 364/300, 364/200 MS File, 364/900 MS FIle, 382/14, 382/15
• US Class Current: 706/11
• CPC Class Codes: G05B 13/028 using expert systems only
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: Y10S 706/906 Process plant
• CPC Class Codes: Y10S 706/914 Process plant

Abstract

An expert system for process control, which permits the inference rules to be revised at any time without requiring the specialized skills of a "knowledge engineer". The inference rules are intitially defined by a domain expert who fills in blank fields in a set of highly constrained substantially natural-language templates. The rule set thus specified is automatically translated to define an operational expert system. Updating can be performed by a domain expert at any time: the set of templates with the data fields as originally entered is redisplayed, so that the domain expert can edit the accessible fields and then command the modified rule set to be automatically translated, to define a modified operational expert system.

Claims

1. An expert system comprising:

a processor connected to receive inputs from a plurality of sources;

a collection of inference rules which are executable by said processor; and

one or more output channels, connected so that said processor provides outputs on said output channels in accordance with inputs received on said input channels;

wherein said processor is also configured to, on command of a user, present said inference rules in a constrained format which is readily understandable by a user who is not necessarily competent in any computer language, and permit said user to alter said executable rules by modifying said rules within said constrained substantially natural language format.

13. The system of claim 1, wherein said inference rules are presented to the user as functional structures which have a substantially natural language format which is readily understandable by a user who is technically skilled in a predetermined art but who is not necessarily competent in any computer language.

14. The system of claim 1, wherein said user-alterable portions of said functional structures appear differently to said user than do other portions of said functional structures.

16. The system of claim 1, wherein said inference rules defines a substantially real-time expert control system.

18. The system of claim 1, wherein said processor is connected to provide control parameter definitions to controllers in a manufacturing process which is operating substantially continuously.

2. A computer-based system for building an expert system, comprising:

rule generation logic, which when activated provides to a user functional structures for no more than three different rule types, wherein said rule generation logic presents said functional structures for all of said rules in a format which is readily understandable by a user who is not necessarily competent in any computer language, and which is not user-alterable except in restricted portions thereof;

and wherein said rule generation logic translates user inputs in accordance with said functional structures into a complete executable set of rules which defines an expert system.

9. The system of claim 2, wherein said rule generation logic selectively:

presents said functional structures for new rules;

or, on command of an authorized user, presents functional structures corresponding to the user input from which a current version of an expert system was generated.

11. The system of claim 2, wherein said rule generation logic, when activated, provides to a user functional structures for rules according to no more than four predetermined types.

17. The system of claim 2, wherein said rule generation logic verifies that each data value referred to by any of said inference rules is well-defined.

28. The method of claim 9, wherein said user-alterable portions of said functional structures appear differently to said user than do other portions thereof.

3. A computer-based system for building an expert system, comprising:

rule generation logic, which when activated provides to a user functional structures for rules according to a limited set of predetermined types,

wherein said rule generator logic presents said functional structures for all of said rules in a format

which is readily understandable by a user who is not necessarily competent in any computer language,

and which is not user-alterable except in restricted portions thereof;

and wherein said rule generation logic translates user inputs in accordance with said functional structures into a complete executable set of rules which defines an expert system.

10. The system of claim 3, wherein said rule generation logic selectively:

presents said functional structures for new rules;

or, on command of an authorized user, presents functional structures corresponding to the user input from which a current version of an expert system was generated.

12. The system of claim 3, wherein said rule generation logic, when activated provides to a user functional structures for rules according to no more than four predetermined types.

15. The system of claim 3, wherein said user-alterable portions of said functional structures appear differently to said user than do other portions of said functional structures.

4. A computer-based method for processing data according to expert knowledge, comprising the steps of:

providing inputs, including numeric inputs, from one or more sources to a processor;

executing in said processor a collection of executable inference rules on said inputs; and

providing outputs on one or more output channels in accordance with inputs received on said input channels and in accordance with said inference rules;

and also, when commanded by an authorized user, displaying said inference rules in a constrained format which is understandable by a user who is not necessarily competent in any computer language, and permitting said user to alter said executable rules by editing said rules within said constrained substantially natural language format.

22. The method of claim 4, wherein said constrained substantially natural language format is readily undeerstandable by a user who is technically skilled in a predetermined art but who is not necessarily competent in any computer language.

29. The method of claim 4, wherein said rules defines a substantially real-time expert control method.

5. A computer-based method for building an expert system, comprising the steps of:

repeatedly displaying to a user at an interactive interface templates for inference rules, wherein said templates for all of said rules are presented in a format which is readily understandable by a user who is not necessarily competent in any computer language, which is not user-alterable except in restricted portions thereof;

translating user inputs in accordance with said templates into an executable set of rules which defines an expert system; and

storing said user inputs, apart from said executable set of rules, in a format such that said user inputs can subsequently be redisplayed to a user for modification, and subsequent user modifications to said inputs can be retranslated to redefine said executable set of rules and thereby redefine said expert system.

23. The method of claim 5, wherein said format is readily understandable by a user who is technically skilled in a predetermined art but who is not necessarily competent in any computer language.

27. The method of claim 5, wherein said user-alterable portions of said templates appear differently to said user than do other portions thereof.

6. A computer-based method for building an expert system, comprising the steps of:

providing to a user functional structures, for rules according to a limited set of predetermined types, in a format

which is readily understandable by a user who is not necessarily competent in any computer language,

and which is not readily user-alterable except in restricted portions thereof;

and translating user inputs in accordance with said functional structures into a complete executable set of rules which defines an expert system.

19. The method of claim 6, wherein said limited set of predetermined rule types consists of no more than four predetermined types.

24. The method of claim 6, wherein said format is readily understandable by a user who is technically skilled in a predetermined art but who is not necessarily competent in any computer language.

30. The method of claim 6, wherein said translating step verifies that each data value referred to by any of said respective rules is well-defined.

7. A computer-based method for building an expert system, comprising the steps of:

providing to a user functional structures, for rules according to a limited set of predetermined types, in a format

which is readily understandable by a user who is not necessarily competent in any computer language,

and which is not readily user-alterable except in restricted portions thereof;

translating user inputs in accordance with said functional structures into a complete executable set of rules which defines an expert system;

and storing

both said executable rule set

and also said user inputs in accordance with said functional structures.

20. The method of claim 7, wherein said limited set of predetermined rule types consists of no more than four predetermined types.

25. The method of claim 7, wherein said format is readily understandable by a user who is technically skilled in a predetermined art but who is not necessarily competent in any computer language.

31. The method of claim 7, wherein said translating step verifies that each data value referred to by any of said respective rules is well-defined.

8. A computer-based method for building an expert system, comprising the steps of:

providing to a user, at an interactive interface, templates embodying functional structures for rules according to a limited set of predetermined types, in a format

which is readily understandable by a user who is not necessarily competent in any computer language,

and which is not readily user-alterable except in restricted portions thereof;

translating user inputs in accordance with said functional structures into a complete executable set of rules which defines an expert system;

and storing

both said executable rule set

and also said user inputs in accordance with said functional structures.

21. The method of claim 8, wherein said limited set of predetermined rule types consists of no more than four predetermined types.

26. The method of claim 8, wherein said format is readily understandable by a user who is technically skilled in a predetermined art but who is not necessarily competent in any computer language.

32. The method of claim 8, wherein said translating step verifies that each data value referred to by any of said respective rules is well-defined.