Problem solving method utilizing emergent clustering and case retrieval

• US 20060167663A1
• Filed: 01/21/2005
• Published: 07/27/2006
• Est. Priority Date: 01/21/2005
• Status: Active Grant

• Current Assignee: Techteam Government Solutions Incorporated (Jacobs Solutions, Inc.), Newvectors, LLC (Stefanini International Holdings Ltd. (Brazil))
• Original Assignee: Sven Brueckner
• Inventors: Brueckner, Sven
• Granted Patent: US 7,318,012 B2
• Time in Patent Office: Days
• US Class Current: 703/2
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

An Adaptive, Any-Time Case Retrieval process combines existing knowledge (emergent clustering, case retrieval with CRN) in a novel and advantageous way. Although ideally suited to the Digital Body Development System (DBDS), the method is applicable to any problem solving environment involving iterative simulations. Given a description of a problem to be solved, a candidate solution is applied to generate a problem solving base case. The description of the problem is modified and the modification is recorded. Candidate solutions are applied to the problem as modified, thereby generating a problem solving event case. These steps are repeated so as to rank the relevance of the cases generated to arrive at an optimal solution to the problem. In the preferred embodiment a case retrieval network (CRN) structure is used in the ranking process.

Claims

1. A method of selecting historical problem/solution cases for a design having a case base, comprising the steps of:

a) creating an arrangement of anchor points that combine observed deviations from functional specifications at predefined measurement points on the design;

b) providing a set of historical problem/solution cases associated with the design;

c) modifying the arrangement of the anchor points and extracting a dynamic problem signature from the new arrangement;

d) ranking the set of historical problem/solution cases according to their similarity to the extracted signature;

e) repeating steps c) and d) until rank scoring is of sufficient quality;

f) selecting a problem/solution case based on its rank score;

g) evaluating the selected problem/solution case based upon divergence from the functional specifications; and

h) repeating steps c) through f) as necessary to identify one or more problem/solution cases applicable to the design.

2. The method of claim 1, wherein the anchor points include additional contextual data describing the problem domain.

3. The method of claim 1, wherein the step of selecting a problem/solution case is carried out probabilistically.

4. The method of claim 1, wherein the step of evaluating a selected case uses a resource-aware adaptive dynamic search infrastructure.

5. The method of claim 1, wherein the step of modifying the arrangement of the anchor points is based upon emergent clustering.

6. The method of claim 5, wherein the anchor points are treated as active clustering agents that relocate according to dynamically computed virtual forces.

7. The method of claim 6, wherein the clustering agents are attracted to other, nearby agents as well as their own anchor point.

8. The method of claim 1, wherein the dynamic problem signature is the set of locations of the clusters of agents at a particular point in time.

9. The method of claim 8, wherein the location of clusters is determined through discrete binning of clustering agent locations.

10. The method of claim 5, wherein the Shannon Entropy of the case selection probabilities changes the forces in the emergent clustering.

11. The method of claim 5, wherein the evaluation of a selected case changes the forces in the emergent clustering.

12. The method of claim 9, wherein the anchor point locations are binned by a pheromone infrastructure (PI) and repeated deposits of digital pheromones at the locations of anchor points.

13. The method of claim 12, wherein signature locations are recorded as a subset of locations of bins in the PI.

14. The method of claim 1, wherein each historical case has a recorded problem signature.

15. The method of claim 1, wherein the historical cases are arranged in a case retrieval network (CRN).

16. The method of claim 15, wherein the CRN includes information entity nodes representing observed locations of clusters of anchor points.

17. The method of claim 15, wherein the CRN is activated by the location of the clusters of anchor points defined in the current problem signature.

18. The method of claim 15, wherein the rank score of each historical case is determined by its current activation in the CRN.

19. The method of claim 1, wherein the probability of selecting a historical case is proportional to its rank score.

20. The method of claim 1, wherein the quality of the rank scoring is the Shannon Entropy of the selection probabilities associated with the cases in the case base.

21. The method of claim 1, wherein the evaluation of a selected case includes a numerical score.