Comparison of TT and MAWG Metadata Properties

[Bailer, Werner]

Dear all,

I have done a review of the metadata elements you propose in the DFXP draft to see how this fits together with our work in the media annotations group (MAWG). MAWG has defined a set of core properties in a recent working draft (http://www.w3.org/TR/2009/WD-mediaont-10-20090618/#core-property-lists). Data types are not yet defined in this draft, this is currently ongoing work.

Here is an attempt to map the elements between the ttm:* and ma:* elements:

- ttm:title: maps to a ma:title instance, it is more general as ma:title can also have qualifiers that specify the type of title (main, secondary, original, ...)

- ttm:desc: 1:1 mapping to ma:description

- ttm:copyright: 1:1 mapping to ma:copyright

- ttm:agent: maps to ma:creator or ma:contributor, ttm:agent has qualifiers (person, character, group, organization, other) that specify the type of agent rather than the role in the creation process (as for ma:creator and ma:contributor)

-- ttm:name: maps to the content of ma:creator/ma:contributor; there are qualifiers for specific name parts or for an unstructured string, similar to the structured/unstructured return values drafted for the MAWG API [note that there is not yet a public WD of the API]

-- ttm:actor: links a character to a real-world agent

To conclude: Most of the ttm elements are covered in the MAWG draft and mapping seems to be feasible. You might want to consider renaming "desc" to "description", as (i) it is the only abbreviated name and (ii) then all elements that map nicely to ma elements would have the same names in both vocabularies.

The only element that is different is the agent element, which not only includes creators and contributors, but also fictional characters. While ttm lacks a way to express the role in contribution, ma lacks a way to express the character impersonated by an actor. Attempting to harmonize this could improve both vocabularies.

Best regards,
Werner

--------------------------------------------------------------------
  Werner Bailer
  Institute of Information Systems & Information Management
  JOANNEUM RESEARCH Forschungsgesellschaft mbH
  Steyrergasse 17, A-8010 Graz, AUSTRIA
 
  phone:  [phone]               mobile: [phone]
  web:    http://iis.joanneum.at            fax: [phone]      
  e-mail: mailto:werner.bailer@...
--------------------------------------------------------------------

 

[Glenn Adams]

Thanks for this useful review. You might find it interesting to review Table K-1 [1] to see the source derivation of the ttm:* element vocabulary. As you can see, it is based in part on SVG (by direct reuse) and MPEG-7 (by conceptual reuse). We have previously had comments about the abbreviation of ttm:desc. I personally don't have a strong feeling either way. We simply used the abbreviation found in SVG, but as you say, it is the only metadata element name that is an abbreviation, although clearly there are other element names which are abbreviations in the remaining DFXP vocabulary.

G.

On Tue, Jun 30, 2009 at 4:29 PM, Bailer, Werner <werner.bailer@...> wrote:

Dear all,

I have done a review of the metadata elements you propose in the DFXP draft to see how this fits together with our work in the media annotations group (MAWG). MAWG has defined a set of core properties in a recent working draft (http://www.w3.org/TR/2009/WD-mediaont-10-20090618/#core-property-lists). Data types are not yet defined in this draft, this is currently ongoing work.

Here is an attempt to map the elements between the ttm:* and ma:* elements:

- ttm:title: maps to a ma:title instance, it is more general as ma:title can also have qualifiers that specify the type of title (main, secondary, original, ...)

- ttm:desc: 1:1 mapping to ma:description

- ttm:copyright: 1:1 mapping to ma:copyright

- ttm:agent: maps to ma:creator or ma:contributor, ttm:agent has qualifiers (person, character, group, organization, other) that specify the type of agent rather than the role in the creation process (as for ma:creator and ma:contributor)

-- ttm:name: maps to the content of ma:creator/ma:contributor; there are qualifiers for specific name parts or for an unstructured string, similar to the structured/unstructured return values drafted for the MAWG API [note that there is not yet a public WD of the API]

-- ttm:actor: links a character to a real-world agent

To conclude: Most of the ttm elements are covered in the MAWG draft and mapping seems to be feasible. You might want to consider renaming "desc" to "description", as (i) it is the only abbreviated name and (ii) then all elements that map nicely to ma elements would have the same names in both vocabularies.

The only element that is different is the agent element, which not only includes creators and contributors, but also fictional characters. While ttm lacks a way to express the role in contribution, ma lacks a way to express the character impersonated by an actor. Attempting to harmonize this could improve both vocabularies.

Best regards,
Werner

--------------------------------------------------------------------
 Werner Bailer
 Institute of Information Systems & Information Management
 JOANNEUM RESEARCH Forschungsgesellschaft mbH
 Steyrergasse 17, A-8010 Graz, AUSTRIA

 phone:  [phone]               mobile: [phone]
 web:    http://iis.joanneum.at            fax: [phone]
 e-mail: mailto:werner.bailer@...
--------------------------------------------------------------------

[Bailer, Werner]

Dear Glenn,

> You might find it interesting
> to review Table K-1 [1] to see the source derivation of the
> ttm:* element vocabulary. As you can see, it is based in part
> on SVG (by direct reuse) and MPEG-7 (by conceptual reuse).

Thanks for this hint. Looking a the table it states that ttm:actor is derived from mpeg7:Creator and ttm:agent from mpeg7:Agent. mpeg7:Agent is quite generic and used in different contexts. If I understand this correctly, ttm:actor[@type='character'] maps to mpeg7:Creator/Character, and  ttm:actor[@type!='character'] maps to mpeg7:Creator/Agent. In addition we know that if an Agent is referenced by a Character, mpeg7:Creator/Role is "actor".

What was the reason for not including an attribute/element for specifying other roles than actor, e.g. director, composer, ... ?

> We
> have previously had comments about the abbreviation of
> ttm:desc. I personally don't have a strong feeling either
> way.

I also do not consider this a major issue, as long as the mapping is clear.

Best regards,
Werner
 

> On Tue, Jun 30, 2009 at 4:29 PM, Bailer, Werner
> <werner.bailer@...> wrote:
>
>
> Dear all,
>
> I have done a review of the metadata elements you
> propose in the DFXP draft to see how this fits together with
> our work in the media annotations group (MAWG). MAWG has
> defined a set of core properties in a recent working draft
> (http://www.w3.org/TR/2009/WD-mediaont-10-20090618/#core-prope

rty-lists). Data types are not yet defined in this draft, this > is currently ongoing work.

>
> Here is an attempt to map the elements between the
> ttm:* and ma:* elements:
>
> - ttm:title: maps to a ma:title instance, it is more
> general as ma:title can also have qualifiers that specify the
> type of title (main, secondary, original, ...)
>
> - ttm:desc: 1:1 mapping to ma:description
>
> - ttm:copyright: 1:1 mapping to ma:copyright
>
> - ttm:agent: maps to ma:creator or ma:contributor,
> ttm:agent has qualifiers (person, character, group,
> organization, other) that specify the type of agent rather
> than the role in the creation process (as for ma:creator and
> ma:contributor)
>
> -- ttm:name: maps to the content of
> ma:creator/ma:contributor; there are qualifiers for specific
> name parts or for an unstructured string, similar to the
> structured/unstructured return values drafted for the MAWG
> API [note that there is not yet a public WD of the API]
>
> -- ttm:actor: links a character to a real-world agent
>
> To conclude: Most of the ttm elements are covered in
> the MAWG draft and mapping seems to be feasible. You might
> want to consider renaming "desc" to "description", as (i) it
> is the only abbreviated name and (ii) then all elements that
> map nicely to ma elements would have the same names in both
> vocabularies.
>
> The only element that is different is the agent
> element, which not only includes creators and contributors,
> but also fictional characters. While ttm lacks a way to
> express the role in contribution, ma lacks a way to express
> the character impersonated by an actor. Attempting to
> harmonize this could improve both vocabularies.
>
> Best regards,
> Werner
>
>
> --------------------------------------------------------------------
> Werner Bailer
> Institute of Information Systems & Information Management
> JOANNEUM RESEARCH Forschungsgesellschaft mbH
> Steyrergasse 17, A-8010 Graz, AUSTRIA
>
> phone:  [phone]               mobile:
> [phone]
> web:    http://iis.joanneum.at            fax: [phone]
> e-mail: mailto:werner.bailer@...
>
> --------------------------------------------------------------------
>
>
>
>
>
>
>

[Philippe Le Hegaret]

On Tue, 2009-06-30 at 13:17 +0200, Bailer, Werner wrote:
> > You might find it interesting
> > to review Table K-1 [1] to see the source derivation of the
> > ttm:* element vocabulary. As you can see, it is based in part
> > on SVG (by direct reuse) and MPEG-7 (by conceptual reuse).
>
> Thanks for this hint. Looking a the table it states that ttm:actor is derived from mpeg7:Creator and ttm:agent from mpeg7:Agent. mpeg7:Agent is quite generic and used in different contexts. If I understand this correctly, ttm:actor[@type='character'] maps to mpeg7:Creator/Character, and  ttm:actor[@type!='character'] maps to mpeg7:Creator/Agent. In addition we know that if an Agent is referenced by a Character, mpeg7:Creator/Role is "actor".
>
> What was the reason for not including an attribute/element for specifying other roles than actor, e.g. director, composer, ... ?

Werner,

there was no requirement to have an element to indicate composer,
director, etc, and that can be handled by private metadata. Thus, we
didn't add them in the specification.

Philippe