Method for controlling a decisional process when pursuing an aim in a specific field of application, such as economical, technical, organizational or similar and system for implementing the method

• US 6,442,438 B1
• Filed: 11/19/1999
• Issued: 08/27/2002
• Est. Priority Date: 04/16/1999
• Status: Expired due to Term

• Current Assignee: Martine Naillon
• Original Assignee: Martine Naillon
• Inventors: Naillon, Martine
• Primary Examiner(s): PATEL, RAMESH B
• Application Number:
  US09/443,371
• Time in Patent Office: 1,012 Days
• Field of Search: 700/28, 700/24, 700/30, 700/31, 700/32, 700/36, 700/49, 700/103, 700/104, 707/3-5, 707/10, 707/100, 707/101, 707/200, 703/6, 706/10-11, 706/14-15, 706/45-50, 706/414, 709/201, 709/202, 709/203, 709/100
• US Class Current: 700/36
• CPC Class Codes: G06N 5/043 Distributed expert systems; Blackboards
• CPC Class Codes: Y10S 706/914 Process plant
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation

Abstract

A method for controlling a decisional process when pursuing an overall aim in a specific field of application, such as economic, technical, organisational or similar, which is automated by the placing in relation, according to rules and strategies predetermined in accordance with the overall aim of at least one knowledge object reflecting a point of view of the field of application and comprising an item of information interpreted as an alarm sign for the decisional process with other knowledge objects reflecting other points of view or other fonctional competencies of the same field of application or of other fields.

Claims

1. A method for controlling a decisional process in an expert system when pursuing an overall aim in a specific field of application, the method including:

establishing, in accordance with an overall aim, a network having multiple levels of data processing agents, said data processing agents being structured in a hierarchy; and

organizing said agents in said network into loops, with an agent of one level constituting a master agent and at least one child agent in a level directly below said master agent and associated with the master agent so that a strategy followed by said master agent is an aim of said child agent, wherein said agents at a lowermost level of the hierarchy are knowledge object creating agents and said agents located in the levels above said object creating agents are knowledge object constructing agents, providing additional knowledge elements to knowledge objects created by said knowledge object creating agents by placing a knowledge object issued from knowledge objects created in relation with other knowledge objects, reflecting other points of view or other functional competencies in the specific field of application or in other fields, each knowledge object being propagated from the lowermost level of the hierarchy to an uppermost level of the hierarchy and being completed by addition of the additional knowledge elements.

2. The method according to claim 1, wherein

said knowledge object creating agent observes data produced in a field of application with which said knowledge creating agent is associated to create the knowledge object when said knowledge object creating agent detects data which may be interpreted as an alarm signal,

the knowledge objects propagate toward the uppermost level of the hierarchy, through knowledge object constructing agents located in the hierarchy between said object creating agents and the uppermost level of the hierarchy, and

said knowledge object constructing agents place knowledge objects received in relation with one another and add to a knowledge object received additional knowledge elements obtained by placing the knowledge objects received into relation with one another.

3. The method according to claims 2, wherein said network of agents manipulates the knowledge objects which become movable knowledge moving inside said network.

4. The method according to claim 2, wherein said network has a hierarchical structure dependent on the overall aim, each of said agents comprising aim and strategy rules and, if appropriate, action plan rules.

5. The method according to claim 4, wherein each child agent can be a master agent of a loop for a lower level of said network, end loops of said network being formed by analysis loops of knowledge objects comprising a knowledge objects analyser master agent and one of said knowledge object creator agents, the knowledge objects being constructed when moving within said network towards the uppermost level of said network.

6. The method according to claim 5, wherein the movable knowledge objects move within said network inside the loops or between the loops.

7. The method according to claim 5, including successively breaking up the decisional process from the overall aim into individual strategies and aims, the strategy of one of said agents at a level n of said network becoming the aim of said agents of a next lower level, n-1, of said network.

8. The method according to claim 6, including breaking up the decisional process into decisional steps defined as cognitive invariants which are propagated as aims and strategies from level to level in said network.

9. The method according to claim 7, wherein said network has a fractal configuration.

10. The method according to claim 1, wherein, for a given application comprising the overall aim, determining structures of the loops by referring to an agent for structuring services of the loop and a knowledge structuring agent for each service that is created.

11. The method according to one of claims 1, including a cognitive agent associated with an analyser agent for creating an alarm sign knowledge object representative of an alarm sign, and a cognitive agent for establishing a context knowledge object of a context in which the alarm sign knowledge object is located.

12. The method according to claim 11, wherein the context constitutes another knowledge object and the alarm sign knowledge object is updated by the context knowledge object.

13. The method according to claim 11, wherein the cognitive agent is a detector producing numerical values.

14. The method according to claim 11, wherein the cognitive agent is a zone detector in a knowledge structure having a concept tree.

15. The method according to claim 14, wherein the cognitive agent is a detector of common characteristics between the alarm sign knowledge object and the context knowledge object.

16. The method according to claim 2, including cognitive agents for creating alarm sign objects and using context in which the alarm sign objects are located to implement a first invariant for creating alarm sign knowledge objects, such as a sign detection process, the first variant pertaining to a first structured knowledge field.

17. The method according to claim 16, including implementing at least one second invariant for validating knowledge objects created during the first invariant, in a second structured knowledge field different from the first structured knowledge field of the first invariant.

18. The method according to claim 17, wherein, for implementing the second invariant, a cognitive creator agent is associated with an analyser agent for detecting, from the second structured knowledge field, knowledge objects of the first invariant, context knowledge objects agent associated with the analyser agent creating, if appropriate, a creator agent for detecting signals concerning the context knowledge objects and a validation cognitive agent detecting correlations constituting points of support for the alarm sign knowledge objects created during the first invariant.

19. The method according to claim 1, including a Man-Machine dialogue in order that a man can modify rules of the method based on the man'"'"'s knowledge.

20. The method according to claim 19, including showing to the man, during the decisional process, results obtained and offering decisional and intervening strategies to the man.

21. A system for implementing the method according to claim 1, including a network having multiple levels of data processing agents having a hierarchical structure wherein a master agent at a level n is associated with child agents at a lower level, n-1, each child agent being a master agent of child agents of a still lower level, n-2, a lowermost level of said network including analyser agents having respective cognitive agents creating knowledge objects.

22. The system according to claim 21, including at least one individual user as an individual knowledge manager.

23. The system according to claim 21, comprising a plurality of said individual knowledge managers including child agents of a master agent, constituting a collective knowledge manager.

24. The system according to claim 23, comprising a plurality of said collective knowledge managers constituting a community knowledge manager.