What use the AstroOntology
Ed Shaya
eshaya at umd.edu
Tue Mar 6 12:18:07 PST 2007
Tony,
Let me just name some use cases for an astronomy ontology off the top
of my head. We have already mentioned a bunch. They all have to do
with making information machine readable and thereby reduce the time and
effort to the scientist.
Proper metadata (ie, more advanced UCDs): columns can be described
better (eg, Radius at SurfaceBrightness=25th mag/sq arcsec in B-band).
Describe resources better (eg, a table of all stars down to V=16th mag
in Globular Clusters within 5 Kpc and with > 10^6 stars). Describe
datacenters better (eg., V and I-band images and color-magnitude tables
and diagrams of galaxies within 5 Mpc and cz < 500).
Hunting for data: Make use of subclasses of query object (ie., when you
ask for class A you get back anything that has been cataloged as a
subclass of A) (eg, query to all relevant data centers for late-type
spiral galaxies with cz< 10000 and are in clusters and that have I-band
and X-ray data and inclination > 45 deg that have AGN within 1 degree.
This should finds an SBc galaxy that is near a Seyfert even though no
mention is made of the term galaxy in the SBc's metadata nor of AGN in
the Seyfert's metadata).
Hunting for operations/transformation: Unit conversions, coordinate
transformations, physical relationships (eg, With a bunch of data on a
cluster, use whatever relevant relationships to provide distance
measurements to this cluster. So use TF-relationship on spirals, use
Dn-sigma on ellipticals, use 10th brightest galaxy, provide distance
measurements of other researchers but exclude Hubble relationship
distances).
Classification: From a catalog of coarsely classified objects add finer
classification (eg, Start with Sloan catalog with just objects
classified as star, galaxy and output A3-pec star, Sd/Irr, etc
classification) (eq, provide all possible classes for this object which
is an outlier on my (V-B, J-V) color-color diagram, include extinction
in the range Av < 2).
Direct Query on Ontology: Ontology can provide summed knowledge of
properties of objects, so one can gain familiarity on objects outside
your expertise (eg, What are the differences between a Globular Cluster
and an Open Cluster?). Or through ontology plus data search (eg., What
is the maximum number of stars in any Open Cluster?)
Please, others should add their favorite use cases as well.
Ed
Tony Linde wrote:
>> What use is has-emission in? If you are looking for those
>> QSOs within a
>> specific range, as in has-emission-in X-ray of 10^35 < Lx < 10^36 erg?
>
> That really is the point I was getting at. AFAIK an ontology cannot store
> the fact that '10^35 < Lx < 10^36 erg', but only the very broad
> 'has-emission-in X-ray'. Is it still useful?
>
> An application, however, might use the AstroOntology (which, I agree, is
> relatively static) allied to a ruleset relating to high emission X-ray
> objects to deliver useful information to the astronomer.
>
> What I was really after, though, was use cases for employing the
> AstroOntology, from which I hoped we could derive the necessary relations
> (beyond the obvious ones such as 'is-a-type-of' and 'is-a-component-of').
>
> T.
>
>> -----Original Message-----
>> From: Ed Shaya [mailto:eshaya at umd.edu]
>> Sent: 06 March 2007 17:01
>> To: Tony Linde
>> Subject: Re: What use the AstroOntology
>>
>>
>>
>> Tony Linde wrote:
>>> So if we use the working hypothesis that the AstroOntology
>> is fixed and is a
>>> complete description of terms used by astronomers and the
>> relations between
>>> those terms, then what relations do we need for which use cases?
>>>
>>> The use case specified in Andreas et al's draft, has the
>> AstroOntology used
>>> to assist users in looking for resources in the registry.
>> These resources
>>> have both fixed keywords and free text used to describe
>> them. The assistance
>>> consists of widening and narrowing the terms used in the
>> registry search. I
>>> would guess that the most important relations in this case are
>>> 'is-a-type-of' and 'is-synonymous-with' and perhaps
>> 'is-a-component-of'. The
>>> synonyms are used to include all the variants of the term
>> sought by the user
>>> in the search. The types (and perhaps components) are used
>> to allow the user
>>> to choose to include wider or narrower types of the term in
>> the search.
>>> Prior to any use of the AstroOntology (or after any
>> updates), the reasoner
>>> can be run to ensure that all the permutations are forced
>> out: so if A
>>> is-a-type-of B is-a-type-of C then A is-a-type-of C is
>> added; and if B'
>>> is-synonymous-with B then A is-a-type-of B' and B'
>> is-a-type-of C is added.
>>> The first question I would ask is, what use cases do we
>> want to satisfy with
>>> the other relations: has-morphology, has-emission-in etc?
>>>
>>> My second question relates more to some confusion of the
>> AstroOntology
>>> overall. We have a list of astronomical objects (BTW
>> wouldn't it be better
>>> to name the top object AstroObject rather than AstrObject?)
>> and putative
>>> relations between them. BUT we are using this (in the draft
>> use case) to
>>> search a registry of resources, not of objects. The
>> metadata describes the
>>> resource and while such metadata may include the type of
>> object picked out
>>> in a catalogue (eg a catalogue of QSOs or of AGNs) and the
>> AstroOntology
>>> would allow the narrowing or widening of the search on this
>> basis, the fact
>>> that the observations were in the X-ray rather than IR is a
>> fact about the
>>> resource not the objects.
>>>
>>> I'm not sure I'm being clear here. Basically, what use is the
>>> 'has-emission-in' relation? Are we hoping that if the user
>> asks some app to
>>> look for QSOs for her, the app can not only suggest looking
>> for AGNs (if QSO
>>> is-a-type-of AGN) but can also suggest looking at other
>> resources which have
>>> X-ray observations (if QSO has-emission-in the X-ray
>> spectrum range)?
>>
>> What use is has-emission in? If you are looking for those
>> QSOs within a
>> specific range, as in has-emission-in X-ray of 10^35 < Lx < 10^36 erg?
>>
>>> If so,
>>> is this useful to the astronomer or is the identification
>> of the QSO so
>>> context driven that no-one would ever just look in other
>> X-ray observations
>>> for QSOs? (apologies for the astro-drivel!)
>> We want to reach the point that no-ONE would ever have to LOOK. The
>> machines someday will deliver exactly what you are interested
>> in. They
>> will correlate with the other thing that you are interested
>> in, and they
>> will tell you the two-point spatial correlation between the
>> two things.
>>> T.
>>>
>>>> -----Original Message-----
>>>> From: owner-semantics at eso.org
>>>> [mailto:owner-semantics at eso.org] On Behalf Of Anita M. S. Richards
>>>> Sent: 06 March 2007 07:19
>>>> To: Ashish Mahabal
>>>> Cc: semantics at ivoa.net
>>>> Subject: Re: What use the AstroOntology
>>>>
>>>>
>>>> On Mon, 5 Mar 2007, Ashish Mahabal wrote:
>>>>
>>> http://www.Taglocity.com Tags: IVOA, semantics
>>>
>
> http://www.Taglocity.com Tags: IVOA, semantics
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: eshaya.vcf
Type: text/x-vcard
Size: 257 bytes
Desc: not available
URL: <http://www.ivoa.net/pipermail/semantics/attachments/20070306/eab383f6/attachment-0001.vcf>
More information about the semantics
mailing list