From sonali at persistent.co.in  Wed Nov  5 04:58:35 2003
From: sonali at persistent.co.in (Sonali A. Kale)
Date: Wed, 5 Nov 2003 18:28:35 +0530
Subject: UCD matching function
Message-ID: <02c401c3a39c$86360a20$6f07a8c0@persistent.co.in>

Hello,

Could someone throw light on the UCD matching 
function ? Will there be separate parsers written for
the matching ? In VOPlot as of now, we use UCDs 
to look for similar columns from another VOTable.
This is a simple string match, but I guess the matching 
function for the new UCD system will be more complex.

Will the matching be provided as a web service ? How
much data will a software have to carry with it for the 
actual matching (if the matching is done by the software tool) ?
I am especially worried about this since VOPlot is an applet 
and an applet with a heavy footprint will not be usable.

I would really appreciate any help on this.

- Thanks,
Sonali
------------------------------------
Sonali A. Kale
Persistent Systems Private Limited
http://www.persistent.co.in
Ph: +91 (20) 567 8900 (ext. 328)
------------------------------------



From roy at cacr.caltech.edu  Wed Nov  5 06:15:05 2003
From: roy at cacr.caltech.edu (Roy Williams)
Date: Wed, 5 Nov 2003 06:15:05 -0800
Subject: UCD matching function
References: <02c401c3a39c$86360a20$6f07a8c0@persistent.co.in>
Message-ID: <072e01c3a3a7$361c34f0$6401a8c0@Ropy>

Sonali

Thank you for your burst of cold reality in this increasingly abstract
discussion. In debating the structure of UCD2, I have been continually
thinking of you and others who build actual software that
"understands" UCD.

Unfortunately, I cannot answer your questions. I could simply say
there will be a web service for doing the matching, but then you will
say who and where and when. I do not know. Can I call here fo
volunteers to make the UCD2 services? How soon?

I am beginning to think that adopting UCD2 will require a lot of hard
work by a lot of people -- data providers as well as consumers -- and
I am not seeing the concrete and immediate benefits.

Perhaps we should start working towards a very sophisticated RDF-based
model ("UCD3") for alpha release in a year or two. But stay with a
cleaned-up UCD1 for the near to medium term.

Comments anyone?
Roy

--------
Caltech Center for Advanced Computing Research
roy at cacr.caltech.edu
626 395 3670

----- Original Message ----- 
From: "Sonali A. Kale" <sonali at persistent.co.in>
To: <ucd at ivoa.net>
Sent: Wednesday, November 05, 2003 4:58 AM
Subject: UCD matching function


> Hello,
>
> Could someone throw light on the UCD matching
> function ? Will there be separate parsers written for
> the matching ? In VOPlot as of now, we use UCDs
> to look for similar columns from another VOTable.
> This is a simple string match, but I guess the matching
> function for the new UCD system will be more complex.
>
> Will the matching be provided as a web service ? How
> much data will a software have to carry with it for the
> actual matching (if the matching is done by the software tool) ?
> I am especially worried about this since VOPlot is an applet
> and an applet with a heavy footprint will not be usable.
>
> I would really appreciate any help on this.
>
> - Thanks,
> Sonali
> ------------------------------------
> Sonali A. Kale
> Persistent Systems Private Limited
> http://www.persistent.co.in
> Ph: +91 (20) 567 8900 (ext. 328)
> ------------------------------------
>
>


From pfo at star.le.ac.uk  Wed Nov  5 07:00:00 2003
From: pfo at star.le.ac.uk (Patricio F. Ortiz)
Date: Wed, 5 Nov 2003 15:00:00 +0000 (GMT)
Subject: UCD matching function
In-Reply-To: <072e01c3a3a7$361c34f0$6401a8c0@Ropy>
Message-ID: <Pine.GSO.4.44L0.0311051452310.12244-100000@sparky.star.le.ac.uk>

On Wed, 5 Nov 2003, Roy Williams wrote:
>
> Perhaps we should start working towards a very sophisticated RDF-based
> model ("UCD3") for alpha release in a year or two. But stay with a
> cleaned-up UCD1 for the near to medium term.
>
> Comments anyone?

Roy,

I'm all for that! Let's make some necessary modifications to UCD1 and start
using them. I went through examining the ERROR 'box', assuming that
each ERROR could be transformed to "error.QUANTITY_UCD". It works most of
the time, but there are some cases where it simply doesn't work. Bottom line is
that in Vizier's metadata there are around 9000 ERROR columns, which
translate into about 570 new error descriptors.

Assuming what we want to do is to "change the labels" from UCD1 to UCD_N,
then we will preserve the degree of comparability. If we start shuffling
columns from one box to another, and merging boxes, then the work will be
much more!!

Cheers,

Patricio
---
Patricio F. Ortiz			pfo at star.le.ac.uk
AstroGrid project
Department of Physics and Astronomy
University of Leicester			Tel: +44 (0)116 252 2015
LE1 7RH, UK


From Edward.J.Shaya.1 at gsfc.nasa.gov  Wed Nov  5 10:39:24 2003
From: Edward.J.Shaya.1 at gsfc.nasa.gov (Ed Shaya)
Date: Wed, 05 Nov 2003 13:39:24 -0500
Subject: UCD matching function
In-Reply-To: <072e01c3a3a7$361c34f0$6401a8c0@Ropy>
References: <02c401c3a39c$86360a20$6f07a8c0@persistent.co.in> <072e01c3a3a7$361c34f0$6401a8c0@Ropy>
Message-ID: <3FA943DC.20007@gsfc.nasa.gov>

Roy,
    This makes sense to me.
     So wherever you need a UCD you can put an
       rdf:id or rdf:resource that identifies a proper descriptive term 
by URL.
         At that URL is the OWL class within the class hierarchy and 
with its
    definition.  There one can define the dimensional analysis of the 
item and
    how it differs from similar terms.
    However, one does not escape the painful reality that each concept 
needs a name
    and so one still needs to come up with agreement on these names.  
The only
   advantage for this aspect is that one can reflect hierarchies in the 
directory
    structure and the URL with the "/" symbol, as in
      http://ivoa.net/OWL/physics/StokesParameters/S0.owl

Ed

Roy Williams wrote:

>Sonali
>
>Thank you for your burst of cold reality in this increasingly abstract
>discussion. In debating the structure of UCD2, I have been continually
>thinking of you and others who build actual software that
>"understands" UCD.
>
>Unfortunately, I cannot answer your questions. I could simply say
>there will be a web service for doing the matching, but then you will
>say who and where and when. I do not know. Can I call here fo
>volunteers to make the UCD2 services? How soon?
>
>I am beginning to think that adopting UCD2 will require a lot of hard
>work by a lot of people -- data providers as well as consumers -- and
>I am not seeing the concrete and immediate benefits.
>
>Perhaps we should start working towards a very sophisticated RDF-based
>model ("UCD3") for alpha release in a year or two. But stay with a
>cleaned-up UCD1 for the near to medium term.
>
>Comments anyone?
>Roy
>
>--------
>Caltech Center for Advanced Computing Research
>roy at cacr.caltech.edu
>626 395 3670
>
>----- Original Message ----- 
>From: "Sonali A. Kale" <sonali at persistent.co.in>
>To: <ucd at ivoa.net>
>Sent: Wednesday, November 05, 2003 4:58 AM
>Subject: UCD matching function
>
>
>  
>
>>Hello,
>>
>>Could someone throw light on the UCD matching
>>function ? Will there be separate parsers written for
>>the matching ? In VOPlot as of now, we use UCDs
>>to look for similar columns from another VOTable.
>>This is a simple string match, but I guess the matching
>>function for the new UCD system will be more complex.
>>
>>Will the matching be provided as a web service ? How
>>much data will a software have to carry with it for the
>>actual matching (if the matching is done by the software tool) ?
>>I am especially worried about this since VOPlot is an applet
>>and an applet with a heavy footprint will not be usable.
>>
>>I would really appreciate any help on this.
>>
>>- Thanks,
>>Sonali
>>------------------------------------
>>Sonali A. Kale
>>Persistent Systems Private Limited
>>http://www.persistent.co.in
>>Ph: +91 (20) 567 8900 (ext. 328)
>>------------------------------------
>>
>>
>>    
>>
>
>  
>


From sonali at persistent.co.in  Wed Nov  5 04:58:35 2003
From: sonali at persistent.co.in (Sonali A. Kale)
Date: Wed, 5 Nov 2003 18:28:35 +0530
Subject: UCD matching function
Message-ID: <02c401c3a39c$86360a20$6f07a8c0@persistent.co.in>

Hello,

Could someone throw light on the UCD matching 
function ? Will there be separate parsers written for
the matching ? In VOPlot as of now, we use UCDs 
to look for similar columns from another VOTable.
This is a simple string match, but I guess the matching 
function for the new UCD system will be more complex.

Will the matching be provided as a web service ? How
much data will a software have to carry with it for the 
actual matching (if the matching is done by the software tool) ?
I am especially worried about this since VOPlot is an applet 
and an applet with a heavy footprint will not be usable.

I would really appreciate any help on this.

- Thanks,
Sonali
------------------------------------
Sonali A. Kale
Persistent Systems Private Limited
http://www.persistent.co.in
Ph: +91 (20) 567 8900 (ext. 328)
------------------------------------



From roy at cacr.caltech.edu  Wed Nov  5 06:15:05 2003
From: roy at cacr.caltech.edu (Roy Williams)
Date: Wed, 5 Nov 2003 06:15:05 -0800
Subject: UCD matching function
References: <02c401c3a39c$86360a20$6f07a8c0@persistent.co.in>
Message-ID: <072e01c3a3a7$361c34f0$6401a8c0@Ropy>

Sonali

Thank you for your burst of cold reality in this increasingly abstract
discussion. In debating the structure of UCD2, I have been continually
thinking of you and others who build actual software that
"understands" UCD.

Unfortunately, I cannot answer your questions. I could simply say
there will be a web service for doing the matching, but then you will
say who and where and when. I do not know. Can I call here fo
volunteers to make the UCD2 services? How soon?

I am beginning to think that adopting UCD2 will require a lot of hard
work by a lot of people -- data providers as well as consumers -- and
I am not seeing the concrete and immediate benefits.

Perhaps we should start working towards a very sophisticated RDF-based
model ("UCD3") for alpha release in a year or two. But stay with a
cleaned-up UCD1 for the near to medium term.

Comments anyone?
Roy

--------
Caltech Center for Advanced Computing Research
roy at cacr.caltech.edu
626 395 3670

----- Original Message ----- 
From: "Sonali A. Kale" <sonali at persistent.co.in>
To: <ucd at ivoa.net>
Sent: Wednesday, November 05, 2003 4:58 AM
Subject: UCD matching function


> Hello,
>
> Could someone throw light on the UCD matching
> function ? Will there be separate parsers written for
> the matching ? In VOPlot as of now, we use UCDs
> to look for similar columns from another VOTable.
> This is a simple string match, but I guess the matching
> function for the new UCD system will be more complex.
>
> Will the matching be provided as a web service ? How
> much data will a software have to carry with it for the
> actual matching (if the matching is done by the software tool) ?
> I am especially worried about this since VOPlot is an applet
> and an applet with a heavy footprint will not be usable.
>
> I would really appreciate any help on this.
>
> - Thanks,
> Sonali
> ------------------------------------
> Sonali A. Kale
> Persistent Systems Private Limited
> http://www.persistent.co.in
> Ph: +91 (20) 567 8900 (ext. 328)
> ------------------------------------
>
>


From pfo at star.le.ac.uk  Wed Nov  5 07:00:00 2003
From: pfo at star.le.ac.uk (Patricio F. Ortiz)
Date: Wed, 5 Nov 2003 15:00:00 +0000 (GMT)
Subject: UCD matching function
In-Reply-To: <072e01c3a3a7$361c34f0$6401a8c0@Ropy>
Message-ID: <Pine.GSO.4.44L0.0311051452310.12244-100000@sparky.star.le.ac.uk>

On Wed, 5 Nov 2003, Roy Williams wrote:
>
> Perhaps we should start working towards a very sophisticated RDF-based
> model ("UCD3") for alpha release in a year or two. But stay with a
> cleaned-up UCD1 for the near to medium term.
>
> Comments anyone?

Roy,

I'm all for that! Let's make some necessary modifications to UCD1 and start
using them. I went through examining the ERROR 'box', assuming that
each ERROR could be transformed to "error.QUANTITY_UCD". It works most of
the time, but there are some cases where it simply doesn't work. Bottom line is
that in Vizier's metadata there are around 9000 ERROR columns, which
translate into about 570 new error descriptors.

Assuming what we want to do is to "change the labels" from UCD1 to UCD_N,
then we will preserve the degree of comparability. If we start shuffling
columns from one box to another, and merging boxes, then the work will be
much more!!

Cheers,

Patricio
---
Patricio F. Ortiz			pfo at star.le.ac.uk
AstroGrid project
Department of Physics and Astronomy
University of Leicester			Tel: +44 (0)116 252 2015
LE1 7RH, UK


From Edward.J.Shaya.1 at gsfc.nasa.gov  Wed Nov  5 10:39:24 2003
From: Edward.J.Shaya.1 at gsfc.nasa.gov (Ed Shaya)
Date: Wed, 05 Nov 2003 13:39:24 -0500
Subject: UCD matching function
In-Reply-To: <072e01c3a3a7$361c34f0$6401a8c0@Ropy>
References: <02c401c3a39c$86360a20$6f07a8c0@persistent.co.in> <072e01c3a3a7$361c34f0$6401a8c0@Ropy>
Message-ID: <3FA943DC.20007@gsfc.nasa.gov>

Roy,
    This makes sense to me.
     So wherever you need a UCD you can put an
       rdf:id or rdf:resource that identifies a proper descriptive term 
by URL.
         At that URL is the OWL class within the class hierarchy and 
with its
    definition.  There one can define the dimensional analysis of the 
item and
    how it differs from similar terms.
    However, one does not escape the painful reality that each concept 
needs a name
    and so one still needs to come up with agreement on these names.  
The only
   advantage for this aspect is that one can reflect hierarchies in the 
directory
    structure and the URL with the "/" symbol, as in
      http://ivoa.net/OWL/physics/StokesParameters/S0.owl

Ed

Roy Williams wrote:

>Sonali
>
>Thank you for your burst of cold reality in this increasingly abstract
>discussion. In debating the structure of UCD2, I have been continually
>thinking of you and others who build actual software that
>"understands" UCD.
>
>Unfortunately, I cannot answer your questions. I could simply say
>there will be a web service for doing the matching, but then you will
>say who and where and when. I do not know. Can I call here fo
>volunteers to make the UCD2 services? How soon?
>
>I am beginning to think that adopting UCD2 will require a lot of hard
>work by a lot of people -- data providers as well as consumers -- and
>I am not seeing the concrete and immediate benefits.
>
>Perhaps we should start working towards a very sophisticated RDF-based
>model ("UCD3") for alpha release in a year or two. But stay with a
>cleaned-up UCD1 for the near to medium term.
>
>Comments anyone?
>Roy
>
>--------
>Caltech Center for Advanced Computing Research
>roy at cacr.caltech.edu
>626 395 3670
>
>----- Original Message ----- 
>From: "Sonali A. Kale" <sonali at persistent.co.in>
>To: <ucd at ivoa.net>
>Sent: Wednesday, November 05, 2003 4:58 AM
>Subject: UCD matching function
>
>
>  
>
>>Hello,
>>
>>Could someone throw light on the UCD matching
>>function ? Will there be separate parsers written for
>>the matching ? In VOPlot as of now, we use UCDs
>>to look for similar columns from another VOTable.
>>This is a simple string match, but I guess the matching
>>function for the new UCD system will be more complex.
>>
>>Will the matching be provided as a web service ? How
>>much data will a software have to carry with it for the
>>actual matching (if the matching is done by the software tool) ?
>>I am especially worried about this since VOPlot is an applet
>>and an applet with a heavy footprint will not be usable.
>>
>>I would really appreciate any help on this.
>>
>>- Thanks,
>>Sonali
>>------------------------------------
>>Sonali A. Kale
>>Persistent Systems Private Limited
>>http://www.persistent.co.in
>>Ph: +91 (20) 567 8900 (ext. 328)
>>------------------------------------
>>
>>
>>    
>>
>
>  
>