Hierarchical Search in SemantEco Support Varied Ontology Design Patterns

1. Hierarchical Search in SemantEco Support Varied Ontology Design Patterns Session: "Semantics for Biodiversity: Interoperability with genomic and ecological semantics" Patrice Seyed, Evan Patton, and Deborah McGuinness (presented by Nathan Wilson)

2. Introduction • Multiple ontology design patterns for modeling taxonomic classification • Vertebrate Taxonomy Ontology (VTO) • Taxonsare represented as classes • ‘IctaluridaeSubClassOfSiluriformes’ • Phenoscape Ontology • Taxa as Individuals • ‘Ictaluridaesubclade_ofSiluriformes’ • Population thinking, inference of up-propagation from descendent species populations to ancestors • ‘contains_cladeo has_member -> has_member’ • A DL description of a clade is propagated up the clade taxonomy

3. SemantEco Modular Framework and the Hierarchical Search Component • Adheres to model-view-controller software architecture pattern • separation between the underlying representation and that which is presented to the user • Allowing support of varied Knowledge Representation design patterns • Supports user interface rendering for navigation along different axes (e.g., generalization, partonomic, taxonomic)

4. SemantEcoModule Framework and the Hierarchical Search Component • SemantEco module designers can provide custom hierarchical search facets. • Enable flexible navigation of resources via their relationships to others, to providing users with multiple paths for finding data. • Leverages JavaScript Trees library (JSTrees) • Each node maps to an RDF Resource, and the selection of a node triggers construction/execution of a SPARQL pattern for rendering immediate children nodes • A SPARQL query pattern is provided by a module designer along the axes of interest for hierarchical navigation, for retrieving a tree’s root and children nodes. • A data-level SPARQL pattern for search using tree selections is ultimately composed in conjunction

5. Flow • Client interface selection maps to REST-ful web request • Server side SPARQL Query executed, results to client as JSON

6. Phenoscape Faceted App

7. SPARQL Query for “Roots” prefix pheno: <http://vocab.phenoscape.org/> select DISTINCT ?child ?parent where { graph <http://phenoscape-example>{ ?child pheno:subclade_of ?parent . FILTER NOT EXISTS { ?parent pheno:subclade_of ?z } } }

8. SPARQL Query for “Children” prefix pheno: <http://vocab.phenoscape.org/> select DISTINCT ?child ?parent where { graph <http://phenoscape-example>{ ?child pheno:subclade_of<Client selection URI>. } }

9. Conclusions • Enables one to develop ontologies and semantic web applications independently • Either conceptualization described above for taxon modeling can be leveraged in the Hierarchical Search Facet component with the appropriate supporting SPARQL queries in place. • This suits our immediate needs for semantic search in SemantEco as a portal and as an architecture, for a semantically-enabled monitoring environment it supports flexible search backed by semantics.

10. References • http://phenoscape.org/wiki/Individual-based_taxonomy • James P. Balhoff, Peter E. Midford, Hilmar Lapp: Integrating Anatomy and Phenotype Ontologies with Taxonomic Hierarchies. International Conference on Biomedical Ontologies, Buffalo, NY 2011