Method and expert system with evolutionary algorithm for valuating an object

• US 8,175,998 B2
• Filed: 04/14/2009
• Issued: 05/08/2012
• Est. Priority Date: 04/14/2008
• Status: Active Grant

• Current Assignee: Inform Institut Fur Operations Research und Management GmbH
• Original Assignee: inform GmbH (SWJ Engineering GmbH)
• Inventors: Weiler, Adrian, Droste, Stefan
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): Chang, Li-Wu
• Application Number:
  US12/423,144
• Publication Number:
  US 20090259614A1
• Time in Patent Office: 1,120 Days
• Field of Search: None
• US Class Current: 706/47
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03

Abstract

A method for valuating an object is disclosed, which includes an expert system that determines a value for a rule from the features of the object, weights the value with a weight from [0, 1] and valuates the object based on the value with a rule editor for defining rules for the valuation with one respective weight from [0, 1]; a rule module for determining a value per rule from the features and a valuation module are also disclosed for valuating the object based on the values, wherein the valuation module weights each value determined for a rule with the weight associated with the rule; and in a simplified manner, the expert system determines temporal progressions of the weight in an evolutionary algorithm, commencing with a start value, and selects a limit of the weight as a new weight, when the weight is converging, and the evolution module determines temporal progressions of at least one weight in an evolutionary algorithm selecting a new weight based on a converging behavior of the characteristics.

Claims

1. A method for valuating an object with a computer system, comprising the steps of:

assigning start value in a range from 0 to 1 to a weight for a rule;

determining a temporal progression of the weight in the range from 0 to 1 with an evolutionary algorithm;

identifying convergence of the weight to a limit value;

selecting the limit value of the weight as a new weight;

determining a value of an object under the rule from features of the object;

weighting the value of the object with the new weight to form a weighted value of the object; and

valuating the object based on the weighted value of the object,

wherein an expert system is implemented in a computer, includes a rule base and the rule is hierarchically combined with additional rules from the rule base,

wherein numerical, categorial or fuzzy conclusions, or a combination thereof, of the additional rules are logically combined to form a premise of the rule.

2. The method of claim 1, further comprising the steps of:

defining a critical time period;

determining if the weight has converged during the critical time period; and

selecting a new weight less than the starting value if the weight has not converged after expiration of the critical time period.

3. The method of claim 1, wherein the rule has a premise including a logical combination of at least two of numerical, categorial and fuzzy terms.

4. A computer system for valuating an object based on features of the object, comprising:

a processor;

a rule editor for defining rules for a valuation;

a rule module for determining a value for each rule from features of an object;

a valuation module for valuating the object based on the determined values for the rules,

wherein the valuation module weights each determined value with a respective weight in a range from 0 to 1 associated with each rule, and

wherein an evolution module configured to execute an evolutionary algorithm, determines a temporal progression of at least one weight and selects a new weight based upon a convergence of the temporal progression,

wherein an expert system is implemented in a computer, includes a rule base and the rule is hierarchically combined with additional rules from the rule base, and

wherein numerical, categorial or fuzzy conclusions, or a combination thereof, of the additional rules are logically combined to form a premise of the rule.

5. The computer system of claim 4, further comprising a display module for displaying the temporal progression.

6. The computer system of claim 4, further comprising a display module for displaying the new weight.

7. The computer system of claim 4, wherein the object of the valuation is a business transaction or a person or a physical object, and the valuation of the object provides an assessment of a risk associated with the business transaction.

8. The computer expert system of claim 7, wherein the business transaction is a financial transaction and the risk is a fraud attempt.