Tracking name patterns in OWL ontologies Ondej vb

Tracking name patterns in OWL ontologies Ondřej Šváb and Vojtěch Svátek The University of Economics The Department of Information and Knowledge Engineering 9/18/2020 Tracking Name Patterns in OWL Ontologies 1

Motivation of the research o o Exploitation of the ‘intentions of ontology designers’ as implicit knowledge They can be seen as patterns n n n 9/18/2020 Name patterns Logical patterns (n-ary relation, classes as values) Their combination Tracking Name Patterns in OWL Ontologies 2

General Goal o o o Understanding of the structure of concepts in OWL ontologies, Detecting modelling errors, Assessing quality of OWL ontologies. 9/18/2020 Tracking Name Patterns in OWL Ontologies 3

Current goal o o name patterns in taxonomic paths in OWL ontologies wrt. our previous goals: understanding of the structure of concepts in OWL ontologies, detecting modelling errors, assessing quality of OWL ontologies. 9/18/2020 Tracking Name Patterns in OWL Ontologies 4

Assumptions for Name Patterns o naming policy the intentions of the designers self-explaining concept names o but, it also depends on domain. . . o o 9/18/2020 Tracking Name Patterns in OWL Ontologies 5

Agenda for the rest of the talk o o o Atomic Name Patterns Complex Name Patterns – incorporated in hypotheses Experiments with real ontologies Conclusions Future work 9/18/2020 Tracking Name Patterns in OWL Ontologies 6

Agenda for the rest of the talk o o o Atomic Name Patterns Complex Name Patterns – incorporated in hypotheses Experiments with real ontologies Conclusions Future work 9/18/2020 Tracking Name Patterns in OWL Ontologies 7

Atomic Name Pattern 1 o o at a token level Token-level relationship: n Prefix extension o n Infix extension o n Written. Document, Written. Legal. Document Postfix extension o 9/18/2020 Written. Document, Hand. Written. Document, Written. Document. With. Comments Tracking Name Patterns in OWL Ontologies 8

Atomic Name Pattern 2 o Identification of the main term n Prefix extension o o n Infix extension o o n Written. Document = main term Written. Legal. Document Postfix extension o o o Written. Document = main term Hand. Written. Document = main term Written. Document. With. Comments Technically, delimiters divide parts of terms n n n 9/18/2020 Underscore (Concept_name) Hyphen (Concept-name) Change of lowercase letter to uppercase letter (Concept. Name) Tracking Name Patterns in OWL Ontologies 9

Agenda for the rest of the talk o o o Atomic Name Patterns Complex Name Patterns – incorporated in hypotheses Experiments with real ontologies Conclusions Future work 9/18/2020 Tracking Name Patterns in OWL Ontologies 10

Hypotheses o o Hypotheses using complex name patterns 1 st hypothesis If the main term in the name of a class and the main term in the multiword name of its immediate subclass do not correspond or agree then it is likely that there is a conceptual incoherence. Correspondence in terms of token-level relationship (name patterns) 9/18/2020 Tracking Name Patterns in OWL Ontologies 11

1 st hypothesis Example of pattern compliance Class: TP ATOMission. Plan Subclasses: o Individual. Location. Reconnaissance. Request. Mission Missile. Weapon. Attack. Mission ‘Plan’ token missing. Other subclasses do have it. 9/18/2020 Tracking Name Patterns in OWL Ontologies 12

1 st hypothesis Example of pattern non-compliance Class: ATOMission Subclasses: o Aircraft. Repositioning FP According to available comments, it means Aircraft. Repositioning. Mission. It is acceptable, though it is not a hyponym of ‘mission’ in Word. Net. 9/18/2020 Tracking Name Patterns in OWL Ontologies 13

3 rd hypothesis Concept with the same main term in their names should not occur in separate taxonomic paths. An example of pattern compliance Three root classes which have token ‘Mission’ as the main term: ATOMission Air. Tanker. Mission Airlift. Mission It can be used as a hint to merge these branches of taxonomies. 9/18/2020 Tracking Name Patterns in OWL Ontologies 14

4 th hypothesis If two taxonomy paths exist such that one contains a class X and its subclass Y, and the other contains a class Z and its subclass W, such that the name of X is a token-level extension of the name of Z, with a different main term, and the name of Y is a tokenlevel extension of the name of W, with a different main term, then both paths should be linked with some property and the name pattern should probably apply for the descendants of Y and W as well. 9/18/2020 Tracking Name Patterns in OWL Ontologies 15

4 th hypothesis Example of pattern compliance: Taxonomic paths: Cargo. Car (X) (Y) Car. Driver Cargo. Car. Driver (Z) (W) Different main terms, but token extensions. Conceptually different, but related. 9/18/2020 Tracking Name Patterns in OWL Ontologies 16

4 th hypothesis An example of pattern compliance Root classes that are a token-level extension: ATOMission Air. Movement. Re. Supply. Mission. . . ATOMission. Plan Air. Tanker. Cell. Mission. Plan. . . It will identify the ‘parallel’ taxonomies of related entities, but they are conceptually different. 9/18/2020 Tracking Name Patterns in OWL Ontologies 17

Agenda for the rest of the talk o o o Atomic Name Patterns Complex Name Patterns – incorporated in hypotheses Experiments with real ontologies Conclusions Future work 9/18/2020 Tracking Name Patterns in OWL Ontologies 18

Experimental condition o o o Verification of the 1 st hypothesis wrt. three OWL ontologies from public repositories Tracking prefix, infix, postfix name patterns in taxonomic paths For identifying coherence of the main terms, Word. Net was employed 9/18/2020 Tracking Name Patterns in OWL Ontologies 19

Conclusions Accuracy quite good in case of ‘ATO MM’ ontology. Coverage of patterns differs wrt. ontology. 9/18/2020 Tracking Name Patterns in OWL Ontologies 20

Future Work o o o Automatic identification of patterns in ontologies and/or non-compliance parts of taxonomies Make formal model for capturing name patterns and hypotheses Focus also on other lexical items in ontologies – labels, comments, property names, domain of property and range of property 9/18/2020 Tracking Name Patterns in OWL Ontologies 21

Future Work o o o Combine name patterns and logical patterns (n-ary relations, classes as values) and their automatic identification Making implicit knowledge explicit via ontology refactoring, transformation It can be a preprocessing step for further tasks like ontology mapping, merging 9/18/2020 Tracking Name Patterns in OWL Ontologies 22

Thank you for your attention! Questions or comments? 9/18/2020 Tracking Name Patterns in OWL Ontologies 23