IP Library Granted Patent US 8,738,636
Granted Patent B2
US 8,738,636 · App. 12/561,839 · Granted May 27, 2014

Ontology alignment with semantic validation

Inventors: Yves Reginald Jean-Mary (Miami, FL); Mansur Kabuka (Miami, FL)
G06F17/30864
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 8,738,636
App. No.
12/561,839
Granted
May 27, 2014
Kind
B2
Abstract

The present invention relates to computer implemented methods and system for determining correspondences between terms in two or more ontologies. The methods and systems are designed to accept as inputs ontologies in Web Ontology Language (OWL) syntax or any other ontology syntax, to calculate a similarity measure between terms in the ontologies, extract an alignment based on this similarity measure, and verify this alignment according to the semantics contained in the ontologies. This process is designed to be executed iteratively until the similarity measures converge, or until another suitable finalization condition is met. The result of these methods and of the systems implementing these methods is an alignment between two or more ontologies establishing semantic correspondences between the terms in the ontologies.

Claims (41)

1. A computer implemented method for aligning ontologies, the method comprising:

receiving two ontologies represented in their syntax;

iteratively performing the following sequence until a finalization condition is met, thereby resulting in calculation of an alignment of the two ontologies:

calculating a weighted average of one or more similarity measures between all pairs of terms, one from each ontology;

extracting a pre-alignment based on the highest similarity between terms;

performing a semantic verification over this pre-alignment to determine correspondences that cannot be semantically verified; and

determining whether an alignment has been reached between the two ontologies.

2. The method of claim 1 , wherein the two ontologies are represented using the Web Ontology Language.

3. The method of claim 1 , wherein a partial alignment between the two ontologies is provided.

4. The method of claim 1 , wherein an alignment is calculated for more than two ontologies, by calculating alignments between each pair of ontologies and then combining the alignments that have been calculated.

5. The method of claim 4 , wherein a partial alignment between the more than two ontologies is provided.

6. The method of claim 1 , wherein a similarity measure between the more than two ontologies is calculated as a weighted average of following similarity measures: a lexical or terminological similarity, a relational similarity, an internal similarity, and an extensional similarity.

7. The method of claim 6 , wherein a partial alignment between the more than two ontologies is provided.

8. A computer implemented method for the semantic verification of an alignment between two or more ontologies, the method comprising:

receiving two ontologies represented in their syntax;

iteratively performing the following sequence until a finalization condition is met, thereby resulting in calculation of an alignment of the two ontologies:

calculating a weighted average of one or more similarity measures between all pairs of terms, one from each ontology;

extracting a pre-alignment based on the highest similarity between terms;

performing a semantic verification over this pre-alignment to determine correspondences that cannot be semantically verified; and

determining whether an alignment has been reached between the two ontologies;

receiving the alignment between the two or more ontologies;

verifying whether each correspondence in the alignment can be verified through other semantic information contained in the two or more ontologies;

producing a list of invalid correspondences as those that could not be verified through semantic information in the two or more ontologies; and

producing a new alignment without invalid correspondences.

9. The method of claim 8 , wherein semantic verification includes a verification of multiple-entity correspondences.

10. The method of claim 8 , wherein semantic verification includes a verification of crisscross correspondences.

11. The method of claim 8 , wherein semantic verification includes a verification of disjointness-subsumption contradictions.

12. The method of claim 8 , wherein semantic verification includes a verification of subsumption incompleteness.

13. The method of claim 8 , wherein semantic verification includes a verification of equivalence incompleteness.

14. The method of claim 8 , wherein semantic verification includes a verification of domain incompleteness.

15. The method of claim 8 , wherein semantic verification includes a verification of range incompleteness.

16. A data processing system for aligning ontologies, the system comprising:

a display device; and

a processor configured to:

receive two or more ontologies represented in their syntax;

iteratively perform the following sequence until a finalization condition is met, thereby resulting in calculation of an alignment of the two ontologies:

calculate a weighted average of one or more similarity measures between pairs of terms of the two or more ontologies;

extract a pre-alignment based on the one or more similarity measures;

perform semantic verification over the pre-alignment;

iteratively execute the calculation of similarity measures, pre-alignment extraction, and semantic verification until a finalization condition is met; and

determining whether an alignment has been reached between the two ontologies.

Assignments (1)
CONFIRMATORY LICENSE Recorded May 6, 2014
From: INFOTECH SOFT, INC.
To: NATIONAL INSTITUTES OF HEALTH (NIH), U.S. DEPT. OF HEALTH AND HUMAN SERVICES (DHHS), U.S. GOVERNMENT
Reel/Frame 032828/0096 →
Continuity (2)
Provisional Application 61098591 · Sep 19, 2008
Related Publication 20100131516A1 · May 27, 2010