RULE CORRELATION TO RULES INPUT ATTRIBUTES ACCORDING TO DISPARATE DISTRIBUTION ANALYSIS

• US 20130103636A1
• Filed: 02/28/2012
• Published: 04/25/2013
• Est. Priority Date: 10/21/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Baudel, Thomas, Van Ham, Frank
• Granted Patent: US 8,825,589 B2
• Time in Patent Office: Days
• US Class Current: 706/52
• CPC Class Codes: G06N 5/025 Extracting rules from data

Abstract

Embodiments of the present invention provide a method, system and computer program product for visualizing rule input attributes with a rule according to disparate rule attribute distributions. In an embodiment of the invention, a method for visualizing rule input attributes with a rule according to disparate rule attribute distributions has been provided. The method includes identifying in response to a selection of a rule for viewing in a rule viewer an input for the selected rule and determining an attribute for the input. However, attributes present in a guard for the selected rule can be excluded. The method also can include computing a global distribution of the attribute irrespective of the identified input and a specific distribution for the identified input. The attribute can be flagged as being correlated with the selected rule when it is determined that a sufficient disparity exists between the global distribution and the specific distribution.

Claims

1. A method for visualizing rule input attributes with a rule according to disparate rule attribute distributions, the method comprising:

identifying in response to a selection of a rule for viewing in a rule viewer executing in memory of a computer, an input for the selected rule;

determining an attribute for the input;

computing a global distribution of the attribute irrespective of the identified input and a specific distribution for the identified input; and

,

flagging the attribute as being correlated with the selected rule responsive to determining a sufficient disparity between the global distribution and the specific distribution.

2. The method of claim 1, further comprising excluding an attribute from consideration present in a guard for the rule.

3. The method of claim 1, wherein a sufficient disparity exists when a probability of the specific distribution being derived from a random sampling of the global distribution is small below a threshold value.

4. The method of claim 3, wherein the probability is determined according to a chi square test of independence.

5-13. -13. (canceled)