Realtionship types

We (AnitaRichards, ElizabethAuden and TonyLinde) have made a first stab at some of the relationship types that would be used in an astronomical ontology. Below are the ones we came up with and a few examples. In each example, the triple should be (subject predicate object) where subject and object are classes and predicate is the relationship type.

Please be aware that we still aren't sure what we're doing with this stuff or where it is leading. The goal is to have a DAML+OIL (or OWL as the new W3C language is to be called: see documents referenced from http://www.w3.org/2001/sw/WebOnt/) representation of Astronomical terms and usages which can dynamically shape a workflow tool - this may be pie in the sky but the process is teaching us a lot about astronomical nomenclature.

Feel free to comment on these ideas or add your own, either here or in the forum (http://forum.astrogrid.org/read.php?TID=181#520).

• is a UCD subclass of (this is a special case in which we'll represent the UCD structure)
  • POS ... UCD_Thing (Thing tends to be the top level object in an ontology)
  • POS_EQ ... POS
  • POS_EQ_DEC ... POS_EQ
  • POS_EQ_DEC_MAIN ... POS_EQ_DEC
  • POS_EQ_RA ... POS_EQ
  • POS_EQ_RA_MAIN ... POS_EQ_RA

• is a type of (see below)
  • Pos ... Measurement
  • Moon ... PlanetaryBody^?
  • Planet ... PlanetaryBody^?
  • PosEq^? ... Position
  • PlanetaryBody^? ... Object

• has a (see below)
  • Planet ... Moon (0..n)
  • SolarSystem^? ... Planet (1..n)
  • SolarSystem^? ... Star (1..n)

• has an instrument (same as 'has a'?)

• is a property of
  • NumberOfMoons^? ... Planet
  • Position ... Object
  • Magnitude ... Star

• is a discipline of
  • RadioAstronomy^? ... ObservationalAstronomy^?
  • ObservationalAstronomy^? ... Astronomy

• is composed of
  • PosEq^? ... RA
  • PosEq^? ... DEC
  • PosEq^? ... Epoch

• is measured by
  • UVFlux ... Detector

• is determined by

• is observed by

• is written by
  • Book ... Person (1..n)

• is published by
  • Book ... PublishingHouse^?

• is disproved by

• is employed by

• is funded by

• is in a group of

• is defined by
  • Position ... RA, DEC, Epoch

• is derived from

• is described by
  • BaryonDensity^? ... InflationTheory^?

• is an effect of
  • GravitationalLensing^? ... Quasar

• has file format

• has compute resource

• has permission

• is modelled by

• contains metadata
  • SolarEventCat^? ... Magnitude

-- TonyLinde - 03 Aug 2002

Should we aim to keep the direction (up or down in level) the same for all relationship types if possible.

Most of the above seem to be up in direction, so maybe instead of is composed of, we should have:

• is a component of (man-made constructs, whether ideas or mechanical things)
  • RA ... PosEq^?
  • DEC ... PosEq^?
  • Epoch ... PosEq^?

and then, instead of has a, two relationship types:

• is a part of (natural things)
  • Planet (1..n) ... SolarSystem^?
  • Star (1..3) ... SolarSystem^?

• is attached to
  • Moon (0..n) ... Planet

from which I guess the inference engine could define reverse relationship type has a from both the above.

-- TonyLinde - 05 Aug 2002

We should probably use the <rdfs:subClassOf> construct where we have defined above the relationship type is a type of. So, in OilEd, define the SubclassOf^? property as the Class which the new class is a type of.

-- TonyLinde - 05 Aug 2002