Agent engine

• US 7,711,670 B2
• Filed: 12/24/2002
• Issued: 05/04/2010
• Est. Priority Date: 11/13/2002
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP AG (SAP SE)
• Inventors: Roediger, Karl Christian (Chris)
• Primary Examiner(s): Sparks; Donald
• Assistant Examiner(s): Tran; Mai T
• Application Number:
  US10/328,855
• Publication Number:
  US 20040098358A1
• Time in Patent Office: 2,688 Days
• Field of Search: 706/46, 717/104, 717/106, 717/163
• US Class Current: 706/46
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models

Abstract

An agent engine includes a definition process, the definition process operable to define a data set associated with an objective, a library storing a set of components, the components comprising at least one of a pre-programmed application, object, algorithm, function, and data set definition, and an agent generator process, the agent generator process operable to define at least one agent that includes at least one component from the library, the at least one generated agent defined to perform a function related to the objective.

Claims

1. An agent engine comprising:

a definition component, the definition component to interactively collect user input to define an objective;

a library storing a plurality of components including any one of executable code, rule-based filters, or object oriented programs; and

an agent generator component implemented by a computer processor, the agent generator component to generate at least one agent that includes at least one of the plurality of components from the library, the at least one generated agent to autonomously perform a function related to the objective and produce at least one output; and

a modification component to automatically modify a rule for the at least one agent based on an evaluation of the at least one output.

2. The agent engine of claim 1, further comprising:

a data feed from a first generated agent to a second generated agent, the data feed to pass data produced by the first generated agent to the second generated agent, the passed data usable by the second generated agent to apply a learning algorithm to one of a rule or a data definition related to the objective.

3. The agent engine of claim 2, further comprising:

a feedback loop to determine from the data feed whether the generated agent may be modified to improve performance.

4. The agent engine of claim 1, wherein the objective is a business objective.

5. The agent engine of claim 4, wherein the objective is an objective related to a business entity.

6. The agent engine of claim 4, wherein at least two agents are generated to perform the business objective.

7. The agent engine of claim 1, further comprises:

a generated agent having at least one of a filter application, a data aggregator, an association application, an analyzer, a modeler, or an actuator.

8. The agent engine of claim 1, further comprises:

a memory system, the memory system to store results produced from the generated agent.

9. The agent engine of claim 1,

wherein the agent generator component defines a key performance indicator having a value or range of values.

10. The agent engine of claim 9, wherein the key performance indicator defines the data set for a generated agent, the data set associated with the objective.

11. The agent engine of claim 1, further comprising:

a data correlation component to determine one of a modification and a reinforcement to at least one of a rule, an action, or a generated agent.

12. The agent engine of claim 1, wherein at least one generated agent includes executable code to modify at least one of a rule or a key performance indicator.

13. The agent engine of claim 12, wherein a modification to the generated agent is based in part on a comparison of results produced during different cycles of execution of the generated agent.

14. The agent engine of claim 1, further comprises:

a fuzzy logic algorithm to evaluate a degree of completion of the generated agent.

15. The agent engine of claim 14, wherein evaluating a degree of completion comprises at least one of comparing different data sets, defining an absolute truth operation, defining a relative truth operation, or evolving a generated agent.

16. The agent engine of claim 1, wherein the definition component is to define a data set associated with the objective.

17. The agent engine of claim 1, wherein the one of the plurality of components included in the agent is selected based on the user input.

18. The agent engine of claim 1, wherein the definition component and the agent generator component are integrated into a definition wizard.

19. The agent engine of claim 1, wherein the agent generator component to generate a plurality of agents to achieve the objective and a first agent in the plurality of agents to route an output of the first agent to an input of a second agent in the plurality of agents.

20. The agent engine of claim 1, further comprising:

a sensor to sense and translate data through a data feed to the at least one agent.

21. The agent engine of claim 1, wherein the modification component to determine a secondary rule in response to the output unsuccessfully meeting the objective when the rule is used.

22. The agent engine of claim 1, wherein the modification component to analyze an effect of modifying the rule.

23. The agent engine of claim 1, wherein the modification component to determine a timeframe to adjust the rule based on a level of importance of the objective and a degree of satisfaction of the objective.

24. The agent engine of claim 1, the modification component to automatically modify the rule based on an empirically successful modification stored in memory.

25. An agent engine architecture system comprising:

a plurality of levels implemented by a computer processor at least one of the plurality of levels to generate at least one or more agents that execute a function to produce a result to satisfy an objective;

at least one of the plurality of levels associated with one of the at least one or more agents,

at least one of the plurality of levels to define the objective,

at least one key performance indicator (KPI) associated with the objective, and rules and parameters associated with the KPI;

at least one of the plurality of levels to filter, analyze, or act based upon a defined data set;

at least one of the plurality of levels to evaluate the result from another level and determine a degree to which the result satisfies the objective;

at least one of the plurality of levels to modify at least one of the KPI, or the at least one of the plurality of levels to learn and determine a modification to at least one of a rule or a generated agent and to automatically modify the rule for the at least one or more agents based on an evaluation of the result; and

at least one definition wizard to receive input to customize an operation within each of the plurality of levels.

26. The system of claim 25, wherein at least one of the plurality of levels comprises a learning system for determining whether modifications to the at least one rule and agent are successful and for modifying the system based on an output of the generated agent.

27. The system of claim 25, further comprising a library of preprogrammed components, the components including at least one of an objective, a KPI, an algorithm, a function or an object.

28. The system of claim 25, further comprising the definition wizard, the definition wizard to generate at least one of a data set or a preprogrammed component for the generated agent.

29. An agent engine system comprising:

a plurality of levels implemented by a computer processor, at least one of the plurality of levels to generate at least one or more agents that execute a function to produce an output to satisfy an objective,

at least one of the plurality of levels associated with one of the at least one or more agents,

a first of the plurality of levels to define the objective customized by interactive user input,

a second the plurality of levels to link at least one of a key performance indicator (KPI) associated with the objective, a rule, or an algorithm associated with the objective,

a third of the plurality of levels to receive the output from one of the plurality of levels and to automatically modify the rule for the at least one or more agents based on an evaluation of the output produced by the at least one or more agents.

30. The system of claim 29, wherein the second of the plurality of levels is further to filter, analyze, or act based upon a defined data set.

31. The system of claim 29, wherein the third of the plurality of levels is further to determine a degree to which the output satisfies the objective, and to modify at least one of the KPI, or the generated agent.

32. The system of claim 29, wherein the third of the plurality of levels is further to determine whether the modification to the one of the rule or the algorithm is successful.

33. The system of claim 29, further comprising a library of preprogrammed components, the components including at least one of an objective, a KPI, an algorithm, a function or an object.

34. The system of claim 29, wherein the first of the plurality of levels further comprises a definition wizard, the definition wizard to generate at least one of a data set or a preprogrammed component for the generated agent.