Expressing position in RDF
eshaya at umd.edu
Wed Oct 15 06:06:46 PDT 2008
Precession must have been explained before somewhere in VOEVENT
discussions, but here is another try at emphasizing it.
Precession rotates the equatorial system at the rate of about 50.3
arcsec per year (approximately 360 degrees divided by 25,700 yr), or 1
degree every 71.6 years. Thus, one year of precession is almost as
large as the distance from the center of the HST/WFPC2 field to the
edge. We avoid vast confusion by stating the positions in a given frame
fixed at a particular epoch (hence J2000, B1950 etc).
In two years GAIA will fly and there will be a vastly superior reference
frame, if all goes well (knock wood). Logically, we will all flock to
this new reference frame which will probably be set at a current epoch
like 2015. So, in not too many years your VOEVENT database will be a
mixed bag of J2000 and G2015. With about 12.5 arcminute disagreements
for the positions of anything with low declination, 10 times larger
than the JWST field of view from center to edge.
Just the facts. You can do what you want with them.
You could, in 2015, replace all positions with G2015 positions, for
instance. But the harder issue is how do you inform all applications
(including someone's one line IDL procedure) that query your database
that this change has been made?
PS - And hopefully there will be an S2020 from SIM.
Matthew Graham wrote:
> RDF is just like XML so questions of production, presentation and
> storage are really quite secondary. It's the representation of
> information that is the primary concern and the issue here is does the
> IVOA have a succinct way of representing celestial positions in RDF. As
> Ed, I think, said, RDF and ontologies are supposed to remove the need
> for context to provide meaning to the information and so how the RDF
> statement is used is also quite secondary. The statement means what it
> says whatever.
> SPARQL is the query language for RDF and is very SQL-like so an
> inequality is just a constraint on the search predicate. The exact
> syntax is dependent on how verbose the RA expression is.
> I would not use a SPARQL query to do intersections, though - we have STC
> and Footprint Services for that. There might an RDF statement that a
> data object has a footprint associated with it and but then the object
> will the URI for the footprint itself.
> On Oct 14, 2008, at 7:35 PM, Roy Williams wrote:
>> I like Doug's question, asking how the RDF will be used: produced,
>> presented, stored etc. Another question in the same genre asks what
>> *queries* will run against this knowledge base. I suspect that "RA"
>> will be used in an arithmetical inequality -- "Give me all sources
>> with RA>240.0", also as part of a cone search or polygon. Is it
>> Sparql, the query language for RDF? How would an inequality be framed
>> as part of a Sparql query?
>> More abstract, we could utilize RDF at a higher level -- not details
>> of coordinate systems, as in this naked quantity "RA". How about the
>> RDF handles "Regions" (of spacetime). With Regions, the questions are
>> no longer arithmetical, but rather boolean choices, about
>> intersection: "Does the Region in which this event lies intersect with
>> the spacetime coverage Region of the Catalina Survey?"
>> California Institute of Technology
>> 626 395 3670
More information about the semantics