Ontology Usage

1
THE US NATIONAL VIRTUAL OBSERVATORY
Ontology Usage

• Edward Shaya, Brian Thomas, Zenping Huang, Peter
  Teuben
• University of Maryland

2
Outline

• Our Project 2-3 years in development, detail
  various potentially valuable uses for the VO,
  tested some sue cases, and developing a framework
  for use in the VO
• Outline
• What can Ontologies do (that is useful to VO) ?
• One use case example we work with
• Our Project Technologies

3
Basic Ontology Properties

• Depicts (Sub)classes, Properties of Classes,
  Domain and Range of Properties, and Instances.
• Subclasses can have restrictions on ranges of
  properties or additional properties.
• Equivalency of Classes, Sameness of Instances,
  Disjointedness, unions,
• Ranges can have Restriction of AllValuesFrom,
  SomeValuesFrom, and Cardinality

4
Obvious Uses

• Explicit statement of terms and relationships
  between terms. Similar to a data dictionary but
  expressed as a directed graph.
• UCD usage connect an element with semantic
  meaning.
• Applications become aware of isa relations SBc
  isa SpiralGalaxy, SpiralGalaxy isa Galaxy,
  TTauri isa Star.
• Aid to translating terminology between interest
  groups.

5
No So Obvious Uses

• Allow for multiple relationships between a pair
  of Classes.
• In XSD, GC/Star means GlobularCluster has Star.
  But in ontology
• GC1 hasForeground Star1
• GC1 hasBackground Star2
• GC1 hasMember Star3
• GC1 hasLineOfSite Star4
• Or
• Region Definedby GC1 has Star4

6
Not So Obvious Uses (2)

• Clearly express Catalogs domain
• Galaxy Milky Way hasMember Messier 3
• GlobularCluster Messier 3
• hasMember CatalogRow
• hasMeasurement ProperMotionpmM3
• Star CatalogRow
• hasMeasurement ProperMotionpmStars
  
• pmStars inInterval pmM3
• This is a way to express that the catalog is one
  of stars that are members of M3 and M3 is in the
  Milky Way and membership is confirmed by proper
  motion.

7
No So Obvious Uses (3)

• In tandem with the proper catalog DM it can
  provides a High Level Query Language
• /MilkyWay//GlobularCluster//Star/hasVelocity//Ampl
  itude/Value gt 35 kmPerSec
• Provides descriptions of transforming between
  different representations of data. Such as from
  vector components to amplitude,direction.
• Ensures valid unit types on each Measurement.
  i.e. Force must have ForceUnits, Mass must have
  MassUnits.

8
Example Use Case

• Use-case Finding the Hubble constant
  observationally.
• User wants distances and redshifts (cz) for
  various galaxy data.

9
Ontology Snipet for Use-case
10
Example Use Case (2)

• Step 1 The Setup - specify the problem
• Galaxy (D, Cz)
• Step 2 Inference engine does initial search
• matches
• Galaxy (D, Cz) SpiralGalaxy (D, Cz)
• Galaxy (Dtf, Cz) SpiralGalaxy (Dtf, Cz)
• Galaxy (Dceph, Cz)

11
Example Use-case (3)

• Step 3 User makes selection of appropriate
  objects. F(for this example we follow only one
  choice)
• SpiralGalaxy (Dtf, Cz)
• Step 4 Inference engine examines functional
  ontology to determine if indirect matches exist.
  Finds that
• Galaxy (Dtf, Cz ) lt-gt Galaxy (W20, Imag, incl)

12
Example Use-case (4)

• Step 5 User presented with the following set of
  semantic data definitions that can be used to
  search the VO repositories.
• SpiralGalaxy (Dtf, Cz )
• SpiralGalaxy (W20, Imag, incl)

13
Ontology Tech we use (right now)

• Jena2 (Java) inference engine
• OWL (W3C standard) for expressing ontologies
• Protege lots of consistency checks,
  visualization, etc to speed ontology construction
• True Grit

14
Summary

• Started work on ontology approach to data
  discovery. Limited (so far) piecemeal success.
• Working on how a unifying framework should look
  like.
• User GUI interface.
• Created initial ontology (at http//archive.astro.
  umd.edu/ont)
• Not exploiting all the possible uses (yet). Could
  be aiding discovery better.