VCal namespace

[Peter Mika-2]

Dear All,

We just have had a very successful VoCamp [1] behind our back, where we
started to discuss among others the issue of representing popular
microformats in RDF. This is a pressing question because large scale
semantic platforms such as Sindice or Yahoo's SearchMonkey would like to
treat microformats at the RDF level  and at the scale at which some of
our systems work there is no possibility for reasoning. Therefore an
agreement on the mapping is required. It doesn't matter as much what
this agreement is, as long as it is an agreement ;)

You can see the outcomes of our current effort at [2]. One of the
problems we spotted was related to VCal. Apparently, there is still
significant confusion as to what the proper namespace for VCal is and
unfortunately the spec at [3] leaves the question in limbo, using either
one or the other namespace at various points.

So my question is: could we deprecate

|http://www.w3.org/2002/12/cal#

|in favor of

http://www.w3.org/2002/12/cal/icaltzd#


and reflecting this change in the documentation at [3] and at [4]?

Thanks,
Peter


[1] http://vocamp.org [2] http://semanticweb.org/wiki/Microformats_in_RDF

[3] http://www.w3.org/TR/rdfcal/

[4] http://www.w3.org/2002/12/cal/

[Dan Connolly]

On Mon, 2008-12-01 at 19:08 +0100, Peter Mika wrote:

> Dear All,
>
> We just have had a very successful VoCamp [1] behind our back, where we
> started to discuss among others the issue of representing popular
> microformats in RDF. This is a pressing question because large scale
> semantic platforms such as Sindice or Yahoo's SearchMonkey would like to
> treat microformats at the RDF level  and at the scale at which some of
> our systems work there is no possibility for reasoning. Therefore an
> agreement on the mapping is required. It doesn't matter as much what
> this agreement is, as long as it is an agreement ;)
>
> You can see the outcomes of our current effort at [2]. One of the
> problems we spotted was related to VCal. Apparently, there is still
> significant confusion as to what the proper namespace for VCal is and
> unfortunately the spec at [3] leaves the question in limbo, using either
> one or the other namespace at various points.
>
> So my question is: could we deprecate
>
> |http://www.w3.org/2002/12/cal#
>
> |in favor of
>
> http://www.w3.org/2002/12/cal/icaltzd#
>
>
> and reflecting this change in the documentation at [3] and at [4]?

Sounds reasonable. Do you have details in mind? Could you suggest
patches?

I haven't looked at this stuff for a while; I should review the
reasons why this wasn't done earlier.

I think the last time I collected my thoughts on all this was:

From: Dan Connolly <connolly@...>
Date: Fri, 14 Apr 2006 01:25:37 -0500
reconsidering timezones in light of hCalendar and CALSIFY
http://lists.w3.org/Archives/Public/www-rdf-calendar/2006Apr/0002.html

I wonder what, if anything, Masahide Kanzaki is currently supporting
in this area. I recall earlier discussion with him about
this namespace...
http://lists.w3.org/Archives/Public/www-rdf-calendar/2004Oct/0004.html

Antoni Mylka wrote up some problems, though I haven't reviewed
them closely.
http://lists.w3.org/Archives/Public/www-rdf-calendar/2007May/0005.html



> Thanks,
> Peter
>
>
> [1] http://vocamp.org [2] http://semanticweb.org/wiki/Microformats_in_RDF
>
> [3] http://www.w3.org/TR/rdfcal/
>
> [4] http://www.w3.org/2002/12/cal/
--
Dan Connolly, W3C http://www.w3.org/People/Connolly/
gpg D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

[Peter Mika-2]

Hi Dan,

Here is what I suggest:

-- Correct mentions of the old namespace |http://www.w3.org/2002/12/cal# 
in [1]:
---- Two occurrences in Section 4
---- Rewrite Section 7 to reflect that this namespace is deprecated

-- Add a visible note to [2] to reflect the change. On the side, I would
also suggest to reorganize the page a bit: it has a lot of old material
(referring to events and chats in 2002!) which have been probably
surpassed by the state-of-the-art.

Cheers,
Peter


[1] http://www.w3.org/TR/rdfcal/
[2] ||http://www.w3.org/2002/12/cal/|

Dan Connolly wrote:

> On Mon, 2008-12-01 at 19:08 +0100, Peter Mika wrote:
>  
>> Dear All,
>>
>> We just have had a very successful VoCamp [1] behind our back, where we
>> started to discuss among others the issue of representing popular
>> microformats in RDF. This is a pressing question because large scale
>> semantic platforms such as Sindice or Yahoo's SearchMonkey would like to
>> treat microformats at the RDF level  and at the scale at which some of
>> our systems work there is no possibility for reasoning. Therefore an
>> agreement on the mapping is required. It doesn't matter as much what
>> this agreement is, as long as it is an agreement ;)
>>
>> You can see the outcomes of our current effort at [2]. One of the
>> problems we spotted was related to VCal. Apparently, there is still
>> significant confusion as to what the proper namespace for VCal is and
>> unfortunately the spec at [3] leaves the question in limbo, using either
>> one or the other namespace at various points.
>>
>> So my question is: could we deprecate
>>
>> |http://www.w3.org/2002/12/cal#
>>
>> |in favor of
>>
>> http://www.w3.org/2002/12/cal/icaltzd#
>>
>>
>> and reflecting this change in the documentation at [3] and at [4]?
>>    
>
> Sounds reasonable. Do you have details in mind? Could you suggest
> patches?
>
> I haven't looked at this stuff for a while; I should review the
> reasons why this wasn't done earlier.
>
> I think the last time I collected my thoughts on all this was:
>
> From: Dan Connolly <connolly@...>
> Date: Fri, 14 Apr 2006 01:25:37 -0500
> reconsidering timezones in light of hCalendar and CALSIFY
> http://lists.w3.org/Archives/Public/www-rdf-calendar/2006Apr/0002.html
>
> I wonder what, if anything, Masahide Kanzaki is currently supporting
> in this area. I recall earlier discussion with him about
> this namespace...
> http://lists.w3.org/Archives/Public/www-rdf-calendar/2004Oct/0004.html
>
> Antoni Mylka wrote up some problems, though I haven't reviewed
> them closely.
> http://lists.w3.org/Archives/Public/www-rdf-calendar/2007May/0005.html
>
>
>
>  
>> Thanks,
>> Peter
>>
>>
>> [1] http://vocamp.org [2] http://semanticweb.org/wiki/Microformats_in_RDF
>>
>> [3] http://www.w3.org/TR/rdfcal/
>>
>> [4] http://www.w3.org/2002/12/cal/
>>    

[KANZAKI Masahide-2]

Hi

I've been using RDFcal for five years in my project, and if the name
space changed this time, it's the second time upset in this short
period. It's very unfortunate for existing projects to have such an
unstable namespace as its building block.

If it is inevitable, please make it complete as soon as possible so
that current project will not waste more resources. If possible,
unchanged namespace is desirable.

(Actually, I'm writing a book, one chapter of which is devoted to RDF
calendar. Stable namespace is very important.)

cheers,

--
@prefix : <http://www.kanzaki.com/ns/sig#> . <> :from [:name
"KANZAKI Masahide"; :nick "masaka"; :email "mkanzaki@..."].

[Peter Mika-2]

Hi Kanzaki,

We're in complete agreement... However, the mistake has been made of not
deciding which namespace to use, and consequently in the past five years
or so people have started using one or the other.

Personally I don't mind if we choose one or the other, but at this
moment Sindice finds 139,000 documents using

http://www.w3.org/2002/12/cal/icaltzd#VEvent

vs. 25 documents using

http://www.w3.org/2002/12/cal#Vevent

The results are most likely heavily biased by the fact that Sindice uses
the former to represent microformats (and we've done the same so far at
Yahoo), but this is how much we have in terms of evidence.

Cheers,
Peter


KANZAKI Masahide wrote:

> Hi
>
> I've been using RDFcal for five years in my project, and if the name
> space changed this time, it's the second time upset in this short
> period. It's very unfortunate for existing projects to have such an
> unstable namespace as its building block.
>
> If it is inevitable, please make it complete as soon as possible so
> that current project will not waste more resources. If possible,
> unchanged namespace is desirable.
>
> (Actually, I'm writing a book, one chapter of which is devoted to RDF
> calendar. Stable namespace is very important.)
>
> cheers,
>
>  

[Dan Brickley-2]

KANZAKI Masahide wrote:

> Hi
>
> I've been using RDFcal for five years in my project, and if the name
> space changed this time, it's the second time upset in this short
> period. It's very unfortunate for existing projects to have such an
> unstable namespace as its building block.
>
> If it is inevitable, please make it complete as soon as possible so
> that current project will not waste more resources. If possible,
> unchanged namespace is desirable.
>
> (Actually, I'm writing a book, one chapter of which is devoted to RDF
> calendar. Stable namespace is very important.)

Hi there! Glad to hear you're writing another book.

I'm glad to hear you consider five years a short time. 2002 seems like
only yesterday to me :)

Peter's main point came from a perceived ambiguity in the rdfcal W3C
Note, about which namespace to use.

"unfortunately the spec at [3] leaves the question in limbo, using
either one or the other namespace at various points",
http://www.w3.org/TR/rdfcal/

The Note lists this as an unresolved issue,
"Note that NY:tz timezone is used as a datatype. Earlier, we used
separate properties for time and timezone, which is initially appealing
but problematic for reasons that are detailed in the
InterpretationProperties  pattern.

     * Objections were raised when this change was made to the original
...2002/12/cal/ical# schema. This design is using a somewhat
experimental2005-03-30 namespace name, ...2002/12/cal/icaltzd#."

...which cites Dan's msg of
http://lists.w3.org/Archives/Public/www-rdf-calendar/2005Mar/0015.html

It's clear from the Note that we don't yet have documented consensus
about the value of these different designs, so it is unclear which
pattern the SearchMonkey folk at Yahoo should be promoting.

In http://lists.w3.org/Archives/Public/www-rdf-calendar/2004Oct/0004.html
  you suggest,

[[
Yes, yes. It's very welcome to roll back the changes in the schema of
current URI, as well as to discuss and develop modified schema with a new
namespace URI.

 >[danc] The tests and conversion tools will migrate to the
 >new schema, I think; I don't think I can afford to
 >keep 2 sets of them around.

That's fine. Keep existing data as is, and move forward.
]]

It seems the existing Note pretty much captures things at this turning
point, which is why it confuses Peter by mentioning two different
namespaces.

I don't see any problem with the original namespace being unchanged and
stable. It should be fine to use, adopt and rely on. The question is
more: do we recommend people use it, or do we recommend people use the
later one with a changed design for timezones? Is anyone beyond Dan
making much use of the later design?

Hmm maybe we can ask the various RDF crawlers about this... take a look
at what has been published in the Web? Of course this wouldn't reflect
private usage, and calendar data is often private or intranet. Perhaps
we can also do Google Code searches or similar?

cheers,

Dan

--
http://danbri.org/

[Antoni Mylka]

Dan Connolly pisze:

> On Mon, 2008-12-01 at 19:08 +0100, Peter Mika wrote:
>> Dear All,
>>
>> We just have had a very successful VoCamp [1] behind our back, where we
>> started to discuss among others the issue of representing popular
>> microformats in RDF. This is a pressing question because large scale
>> semantic platforms such as Sindice or Yahoo's SearchMonkey would like to
>> treat microformats at the RDF level  and at the scale at which some of
>> our systems work there is no possibility for reasoning. Therefore an
>> agreement on the mapping is required. It doesn't matter as much what
>> this agreement is, as long as it is an agreement ;)
>>
>> You can see the outcomes of our current effort at [2]. One of the
>> problems we spotted was related to VCal. Apparently, there is still
>> significant confusion as to what the proper namespace for VCal is and
>> unfortunately the spec at [3] leaves the question in limbo, using either
>> one or the other namespace at various points.
>>
>> So my question is: could we deprecate
>>
>> |http://www.w3.org/2002/12/cal#
>>
>> |in favor of
>>
>> http://www.w3.org/2002/12/cal/icaltzd#
>>
>>
>> and reflecting this change in the documentation at [3] and at [4]?
>
> Sounds reasonable. Do you have details in mind? Could you suggest
> patches?
>
> I haven't looked at this stuff for a while; I should review the
> reasons why this wasn't done earlier.
>
> I think the last time I collected my thoughts on all this was:
>
> From: Dan Connolly <connolly@...>
> Date: Fri, 14 Apr 2006 01:25:37 -0500
> reconsidering timezones in light of hCalendar and CALSIFY
> http://lists.w3.org/Archives/Public/www-rdf-calendar/2006Apr/0002.html
>
> I wonder what, if anything, Masahide Kanzaki is currently supporting
> in this area. I recall earlier discussion with him about
> this namespace...
> http://lists.w3.org/Archives/Public/www-rdf-calendar/2004Oct/0004.html
>
> Antoni Mylka wrote up some problems, though I haven't reviewed
> them closely.
> http://lists.w3.org/Archives/Public/www-rdf-calendar/2007May/0005.html
>

Since my post in May 2007, the Nepomuk Calendar Ontology has been posted
on a server where it is available at its proper namespace. Some
decisions we made come from the Nepomuk background. I'm not saying that
it should be used as it is, but the problems I listed at [1] might be
worth a look.

[1]
http://www.semanticdesktop.org/ontologies/2007/04/02/ncal/#sec-drawbacks

Antoni Mylka
antoni.mylka@...

[Peter Mika-2]

Hi Dan,

Our emails crossed each other but I completely agree:

>
> I don't see any problem with the original namespace being unchanged
> and stable. It should be fine to use, adopt and rely on. The question
> is more: do we recommend people use it, or do we recommend people use
> the later one with a changed design for timezones? Is anyone beyond
> Dan making much use of the later design?

Exactly: given that at the scale we operate we can not afford complex
reasoning to  figure out the relationship between two ontologies (even
if they are 99% same and they differ in a single axiom). So without an
agreement on URIs there will be no interoperability.

Good news is: it's not too late... we are still a small group and there
is room to change things. We can change on our side, Sindice can change
on their side... but the window of opportunity is closing. So the
question: if the Semantic Web is to start today, what namespace should
people use to represent hCalendar and other calendar information in RDF?

Cheers,
Peter

[Dan Brickley-2]

Peter Mika wrote:

>
> Hi Kanzaki,
>
> We're in complete agreement... However, the mistake has been made of not
> deciding which namespace to use, and consequently in the past five years
> or so people have started using one or the other.
>
> Personally I don't mind if we choose one or the other, but at this
> moment Sindice finds 139,000 documents using
>
> http://www.w3.org/2002/12/cal/icaltzd#VEvent
>
> vs. 25 documents using
>
> http://www.w3.org/2002/12/cal#Vevent
>
> The results are most likely heavily biased by the fact that Sindice uses
> the former to represent microformats (and we've done the same so far at
> Yahoo), but this is how much we have in terms of evidence.

It's definitely good to ground this in stats. Can you run a different
query that distinguishes the microformat-converted piece from the rest?

Last time I was at DERI Galway I had some discussions w/ SWSE folks and
we made a prototype, see sample Google doc (w/ flash vizualisation) at
http://spreadsheets.google.com/ccc?key=phsHybBDW1ySxupr8xfL4Zw&hl=en ...
which tries to take into account the spread of vocabulary use across
sites, as well as the raw number of documents / triples. In this case
it's clear such information is critical to decision making: 139k
documents using the icaltzd sounds like a huge endorsement for that
work, ... yet if it all mostly from 5 lines of easily-changed Perl
running on one system, rather than documents out there in the wild.

Here are some related Google Code Search results:

http://www.google.com/codesearch?q=%22http%3A%2F%2Fwww.w3.org%2F2002%2F12%2Fcal%2Ficaltzd%23%22&hl=en&btnG=Search+Code

ie. "http://www.w3.org/2002/12/cal/icaltzd#" ... 142 hits.

Contrasting with only 10 for "http://www.w3.org/2002/12/cal#" .

While 142 is > 10, neither number seems to indicate massively widespread
adoption.

For comparisons:

"http://www.w3.org/1999/02/22-rdf-syntax-ns#" - ~72500
"http://purl.org/dc/elements/1.1/" - ~51500
"http://www.w3.org/2000/01/rdf-schema#" - ~15000
"http://xmlns.com/foaf/0.1/" - ~6000
"http://www.w3.org/2002/07/owl#" - ~6000
"http://purl.org/dc/elements/1.0/" - ~1000
"http://purl.org/dc/terms/" - ~3000
"http://xmlns.com/wordnet/1.6/" - 545
"http://xmlns.com/wot/0.1/"  - 244
"http://www.w3.org/2003/06/sw-vocab-status/ns#" - 160

However "http://rdfs.org/sioc/ns#" only finds 172, which suprises me
given the amount of activity around SIOC and hence makes me cautious of
this approach. Still, the more evidence we can gather the better, and I
think such searches (maybe with a bit of refinement) potentially very
illuminating. It's also worth stressing that these crude metrics would
count equally a line of code in some abandoned test script, versus a
line in a hugely adopted codebase (eg. drupal, livejournal). So clicking
through those results and skimming the detail is probably also important.

cheers,

Dan

--
http://danbri.org/

[KANZAKI Masahide-2]

Oops, sorry, I thought the opposite:

(i.e. I misunderstood that you proposed to use ...2002/12/cal#,
instead of ...2002/12/cal/icaltzd#).

As Danbri mentioned, I've used another original URI
(...2002/12/cal/ical#) for a long (!) time, and recently moved to
...2002/12/cal/icaltzd#, so upset by one more change...

Some problems may still remain, but in my case, it is welcome to make
the namespace URI clear and stable.

cheers,

--
@prefix : <http://www.kanzaki.com/ns/sig#> . <> :from [:name
"KANZAKI Masahide"; :nick "masaka"; :email "mkanzaki@..."].

[Richard Cyganiak-3]

On 2 Dec 2008, at 11:51, Peter Mika wrote:

> Personally I don't mind if we choose one or the other, but at this  
> moment Sindice finds 139,000 documents using
>
> http://www.w3.org/2002/12/cal/icaltzd#VEvent
>
> vs. 25 documents using
>
> http://www.w3.org/2002/12/cal#Vevent
>
> The results are most likely heavily biased by the fact that Sindice  
> uses the former to represent microformats (and we've done the same  
> so far at Yahoo), but this is how much we have in terms of evidence.

Oops -- after excluding microformat-producing sites (using Sindice's "-
domain:www.xyz.com" syntax), only some 20 documents are left using the  
icaltzd namespace. So, both namespaces are equally unpopular.

Not sure what we should do.

Richard

>
>
> Cheers,
> Peter
>
>
> KANZAKI Masahide wrote:
>> Hi
>>
>> I've been using RDFcal for five years in my project, and if the name
>> space changed this time, it's the second time upset in this short
>> period. It's very unfortunate for existing projects to have such an
>> unstable namespace as its building block.
>>
>> If it is inevitable, please make it complete as soon as possible so
>> that current project will not waste more resources. If possible,
>> unchanged namespace is desirable.
>>
>> (Actually, I'm writing a book, one chapter of which is devoted to RDF
>> calendar. Stable namespace is very important.)
>>
>> cheers,
>>
>>

[Dan Brickley-2]

KANZAKI Masahide wrote:

> Oops, sorry, I thought the opposite:
>
> (i.e. I misunderstood that you proposed to use ...2002/12/cal#,
> instead of ...2002/12/cal/icaltzd#).
>
> As Danbri mentioned, I've used another original URI
> (...2002/12/cal/ical#) for a long (!) time, and recently moved to
> ...2002/12/cal/icaltzd#, so upset by one more change...
>
> Some problems may still remain, but in my case, it is welcome to make
> the namespace URI clear and stable.

Ah, I'm glad this was a misunderstanding :)

After this migration, do you have any new insights into the tradeoffs
between the two designs? If you're aware of problems, it seems that we
still could address them since none of this has massive adoption yet.

cheers,

Dan

--
http://danbri.org/

[KANZAKI Masahide-2]

Hi, eh, it's last week when we met last time ;-)

2008/12/2 Dan Brickley <danbri@...>:
> Ah, I'm glad this was a misunderstanding :)
>
> After this migration, do you have any new insights into the tradeoffs
> between the two designs? If you're aware of problems, it seems that we still
> could address them since none of this has massive adoption yet.

Actually, I guess Peter's proposal is not a migration of two designs,
but to clarify the namespace URI for the current design, i.e.
...2002/12/cal# and ...2002/12/cal/icaltzd# use the same date-time
model.

I still keep some amount of documents with original namespace, but I
use separate scripts and stylesheets for them. Some simple rules will
help to migrate them into new model.


btw, I also write a chapter that introduces RDF vcard (2006 version),
whose namespace is another headache...
--
@prefix : <http://www.kanzaki.com/ns/sig#> . <> :from [:name
"KANZAKI Masahide"; :nick "masaka"; :email "mkanzaki@..."].

[Dan Connolly]

On Tue, 2008-12-02 at 13:04 +0100, Peter Mika wrote:
> Hi Dan,

Hi... from a different Dan...

> [...] So the
> question: if the Semantic Web is to start today, what namespace should
> people use to represent hCalendar and other calendar information in RDF?

I suggest
 http://www.w3.org/ns/cal
backed by a W3C XG (which just needs 3 W3C member orgs to start).

I think the test suite is critical too. Note the test suite
(and surrounding toolset) currently uses the icaltzd namespace.

See: URIs for W3C Namespaces
  http://www.w3.org/2005/07/13-nsuri

and

 How to Form an XG
 http://www.w3.org/2005/Incubator/how-to.html


p.s. sorry to be brief... this RDF stuff isn't really my day job
any more... struggling to squeeze this in...

--
Dan Connolly, W3C http://www.w3.org/People/Connolly/
gpg D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

[Peter Mika-2]

I would be happy with a new namespace as well... but then we have to
stick to it. I agree with Section 7 of the document: having to choose
between two evils, I would rather have a changing semantics than a
changing URI. And I think this is why these days people don't encode
creation dates in URIs anymore...

Cheers,
Peter



Dan Connolly wrote:

> On Tue, 2008-12-02 at 13:04 +0100, Peter Mika wrote:
>  
>> Hi Dan,
>>    
>
> Hi... from a different Dan...
>
>  
>> [...] So the
>> question: if the Semantic Web is to start today, what namespace should
>> people use to represent hCalendar and other calendar information in RDF?
>>    
>
> I suggest
>  http://www.w3.org/ns/cal
> backed by a W3C XG (which just needs 3 W3C member orgs to start).
>
> I think the test suite is critical too. Note the test suite
> (and surrounding toolset) currently uses the icaltzd namespace.
>
> See: URIs for W3C Namespaces
>   http://www.w3.org/2005/07/13-nsuri
>
> and
>
>  How to Form an XG
>  http://www.w3.org/2005/Incubator/how-to.html
>
>
> p.s. sorry to be brief... this RDF stuff isn't really my day job
> any more... struggling to squeeze this in...
>
>