OASIS Open Document Format for Office Applications (OpenDocument) TC

After the review of our RDF metadata feature by the inventors of RDF, 
the W3C Semantic Web Interest Group (SWIG), two weeks ago at the TPAC 
2008, I received a lot of positive feed-back.

Nevertheless even good features have potential of improvement, therefore 
I have three subjects I am working on to bring the ODF RDF feature to 
close perfection, driven by the feed-back from SWIG and Tim Berners-Lee 
especially.

Here the rough descriptions of the three topics, a detailed proposal 
will follow:

1)
We are using RDFa attriubtes and the W3C RDFa spec [1] has received 
Recommendation status less than 3 weeks ago.
The usage of the four RDFa elements can now be done accordingly to the 
W3C spec. Even with XHTML namespace due to the XHTML modularization 
statement that all no namespace XHTML attributes can rather use the 
XHTML namespace [2].
The usage of the attribute definitions from the original spec will give 
us furthermore the abbreviation mechanism as compact URIs (CURI), which 
can help a lot in large spreadsheets.
Other RDFa features, e.g. Chaining [3], are rather meant for DOM 
applications and will not apply to ODF applications.

2)
Although ODF applications do not need this for themselves, we should 
help other none-ODF RDF applications to receive the RDF graphs from our 
package, to let ODF become successful.
For this reason I will work on two XSLT transformations (according to 
the W3C GRDDL standard [4]) that will help foreign users/crawlers to 
extract RDF data from ODF.
My basic idea is that styles.xml and content.xml will receive a 
reference to one XSLT stylesheet based on [5] to extract RDFa and 
another XSLT for the meta.xml, making RDFa statements from every 
sub-element of office:meta (e.g. 

On Mon, Nov 3, 2008 at 12:35 PM, Svante Schubert

Bruce D'Arcus wrote:
> On Mon, Nov 3, 2008 at 12:35 PM, Svante Schubert
> 

Bruce D'Arcus wrote:
> On Mon, Nov 3, 2008 at 12:35 PM, Svante Schubert
> 

On Mon, Nov 3, 2008 at 12:35 PM, Svante Schubert

Dear TC members,

I have divided the changes mentioned below into four minor proposals and 
added them to our ODF 1.2 task list [1].
The proposals will be finalized till 10th of December including all 
suggested wordings.

Regards,
Svante

[1]  http://wiki.oasis-open.org/office/OpenDocument_v1.2_tasks

Svante Schubert wrote:
> After the review of our RDF metadata feature by the inventors of RDF, 
> the W3C Semantic Web Interest Group (SWIG), two weeks ago at the TPAC 
> 2008, I received a lot of positive feed-back.
>
> Nevertheless even good features have potential of improvement, 
> therefore I have three subjects I am working on to bring the ODF RDF 
> feature to close perfection, driven by the feed-back from SWIG and Tim 
> Berners-Lee especially.
>
> Here the rough descriptions of the three topics, a detailed proposal 
> will follow:
>
> 1)
> We are using RDFa attriubtes and the W3C RDFa spec [1] has received 
> Recommendation status less than 3 weeks ago.
> The usage of the four RDFa elements can now be done accordingly to the 
> W3C spec. Even with XHTML namespace due to the XHTML modularization 
> statement that all no namespace XHTML attributes can rather use the 
> XHTML namespace [2].
> The usage of the attribute definitions from the original spec will 
> give us furthermore the abbreviation mechanism as compact URIs (CURI), 
> which can help a lot in large spreadsheets.
> Other RDFa features, e.g. Chaining [3], are rather meant for DOM 
> applications and will not apply to ODF applications.
>
> 2)
> Although ODF applications do not need this for themselves, we should 
> help other none-ODF RDF applications to receive the RDF graphs from 
> our package, to let ODF become successful.
> For this reason I will work on two XSLT transformations (according to 
> the W3C GRDDL standard [4]) that will help foreign users/crawlers to 
> extract RDF data from ODF.
> My basic idea is that styles.xml and content.xml will receive a 
> reference to one XSLT stylesheet based on [5] to extract RDFa and 
> another XSLT for the meta.xml, making RDFa statements from every 
> sub-element of office:meta (e.g. 

Now that IS 29500 is available, you might want to look at applying the Part
and Pack URIs of the Open Packaging Conventions for references in and out
and around in ODF packages.  (Actually, the pack: URI scheme is worth
looking at with regard to ODF 1.2 more broadly, since it is already
available for use via the anyURI type according to Annex B of IS 29500-2).

I am thinking specifically of the IS 29500-2:2008 section 9, Package Model,
with subsections 9.1 Parts and 9.2 Part Addressing.  There is no requirement
to support OPC Relationships for ODF, and some other OPC portions can be
avoided as well (e.g., fragments).  I imagine that part URIs can be used
quite nicely in RDF, and the fit into the metadata extensions should work
quite well.  Have the W3C folk looked at this? 

 - Dennis

PS: There is also substantial work on physical packaging and mapping to Zip
that would be useful for ODF in terms of clarity and dealing with the
complexity of URLs whose paths pass through a variety of resolution regimes
(i.e., through a Zip to a file system to something else).  There is also
useful nomenclature In this regard, section 10 has valuable material,
especially with the way it is rigorous around mapping to a Zip archive in
section 10.2.  Annexes A-C are also very useful and we could avoid having to
duplicate that level of work (although there might be exceptions for ODF).
The main difference for ODF packages is the use of the special MIMETYPE item
and the absence of OPC Relationships (but having the RDF-based extensions).

PPS: Although I have wondered about this ever since I saw the early OPC and
ODF specifications, the topic fell off my radar for anything to do with ODF
1.2 until the connection with RDF usage came to mind.  Consequently, I am
going to add consideration of this to the ODF 1.2 task list [1]

PPPS: IS 29500-2:2008 is available as a free download for individual use.
See 
http://nfoworks.org/diary/2008/11/isoiec-295002008-ooxml-standard.htm

 



                                                

					
                                                    

        
                                                
				
                                                

                                                    
                                                
                                                
                                            

Hello Dennis,

Thanks for the feed-back. The package scheme is only one possible 
solution to create an URI to resolve sub-content from a package (like an 
OpenDocument package).
The W3C created yesterday a new mailing list regarding the topic of 
package URI, where I just posted different scenarios and solutions for 
package URIs [1].

The pack URL is no solution for us, because it is still only a 
provisional schema at IANA [2].
As there is no package scheme standardized, I would suggest to await the 
results of the W3C list.

Due to large amount of proposals, I hereby withdraw my fragment 
identifier proposal for ODF 1.2 and postpone it to the successor.
You might want to do the same for your proposal as the scheme is 
provisional, Dennis.

To show you the advantage of the fragment identifier approach over a 
package URI, please let me give you the following preview of the 
fragment proposal:
I will propose the current syntax [3] for all ODF media types and 
furthermore add a mime dependent syntax to provide higher usability.
For instance for an ODF spreadsheet document a user might create an URI 
with the following fragment identifier:

The sub-content of the spreadsheet
    http://www.oasis-open.org/test.ods
might be resolved in the following order by an ODF aware application:

1) Abbreviation to an xml:id of the /content.xml
        http://www.oasis-open.org/test.ods#someID

2) Cell Range of spreadsheet
        http://www.oasis-open.org/test.ods#mysheet.B5:C7

3) Cell of spreadsheet
        http://www.oasis-open.org/test.ods#mysheet.B5  

4) Named Range
        http://www.oasis-open.org/test.ods#someName

5) DataBase Range
        http://www.oasis-open.org/test.ods#someName

6) Number as Line number of spreadsheet
        http://www.oasis-open.org/test.ods#mysheet.42

7) Sheetname
        http://www.oasis-open.org/test.ods#some%20sheet

8) Name of graphic (e.g. chart)
        http://www.oasis-open.org/test.ods#myGraph

Imagine the power for the user to reference a subset of any ODS document 
in the web, even without having write permissions to add an ID!

But that is meant only as a preview to show the advantage over a package 
scheme. The fragment identifier proposal should be discussed after 
finishing ODF 1.2.

Regards,
Svante

[1] http://lists.w3.org/Archives/Public/public-pkg-uri-scheme/
[2] http://www.iana.org/assignments/uri-schemes/prov/pack (cp. 2. Status)
[3] 
http://wiki.oasis-open.org/office/Change_Proposal_for_ODF_1.2_using_URL_fragment_identifiers_for_ODF_media_types 



Dennis E. Hamilton wrote:
> Now that IS 29500 is available, you might want to look at applying the Part
> and Pack URIs of the Open Packaging Conventions for references in and out
> and around in ODF packages.  (Actually, the pack: URI scheme is worth
> looking at with regard to ODF 1.2 more broadly, since it is already
> available for use via the anyURI type according to Annex B of IS 29500-2).
>
> I am thinking specifically of the IS 29500-2:2008 section 9, Package Model,
> with subsections 9.1 Parts and 9.2 Part Addressing.  There is no requirement
> to support OPC Relationships for ODF, and some other OPC portions can be
> avoided as well (e.g., fragments).  I imagine that part URIs can be used
> quite nicely in RDF, and the fit into the metadata extensions should work
> quite well.  Have the W3C folk looked at this? 
>
>  - Dennis
>
> PS: There is also substantial work on physical packaging and mapping to Zip
> that would be useful for ODF in terms of clarity and dealing with the
> complexity of URLs whose paths pass through a variety of resolution regimes
> (i.e., through a Zip to a file system to something else).  There is also
> useful nomenclature In this regard, section 10 has valuable material,
> especially with the way it is rigorous around mapping to a Zip archive in
> section 10.2.  Annexes A-C are also very useful and we could avoid having to
> duplicate that level of work (although there might be exceptions for ODF).
> The main difference for ODF packages is the use of the special MIMETYPE item
> and the absence of OPC Relationships (but having the RDF-based extensions).
>
> PPS: Although I have wondered about this ever since I saw the early OPC and
> ODF specifications, the topic fell off my radar for anything to do with ODF
> 1.2 until the connection with RDF usage came to mind.  Consequently, I am
> going to add consideration of this to the ODF 1.2 task list [1]
>
> PPPS: IS 29500-2:2008 is available as a free download for individual use.
> See 
> http://nfoworks.org/diary/2008/11/isoiec-295002008-ooxml-standard.htm
>   
>  
>
>