Structures for the publication of TIPs

Message
Author
KeithWillshaw
Posts: 77
Joined: Tue May 15, 2012 8:48 am

Re: Structures for the publication of TIPs

#11 Post by KeithWillshaw »

Andrew.Prosser wrote:Notes from todays discussion 20-06-13...... just working out how to attach a file....
Posted on behalf of Andrew
Attachments
TIPs workshop discussion 20-06-13.zip
Posted on behalf of Andrew
(2.62 KiB) Downloaded 678 times

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Structures for the publication of TIPs

#12 Post by vvagr »

Andrew,

Further questions about your format suggestions:

1. It is fortunate that for IIP templates all templates have unique global role identifiers. As we've discussed elsewhere on this forum, it is not always the case and other template sets may reuse roles for many templates. Therefore it will be necessary to refer to particular templates in mapping descriptions, not only to roles. As I can see now, such reference is not included in your examples.

2. As well as I understand the pattern to templates mapping, there are 3 possible parts to describe:

- Pattern parameter is mapped to template role.
- Template role is occupied by a reference data item.
- Two (or more) roles in two (or more) templates are occupied by the same item which belongs to a predefined reference data class but is not explicitly instantiated. For example, in a Coating Color TIP two templates have a member of EXTERNAL COATING class occupying two roles. Keith is describing another such example above.

If I'm not mistaken, your example so far doesn't cover this third possible part? Is it possible to express such correspondence in RDF/OWL?

Andrew.Prosser

Re: Structures for the publication of TIPs

#13 Post by Andrew.Prosser »

1. It is fortunate that for IIP templates all templates have unique global role identifiers. As we've discussed elsewhere on this forum, it is not always the case and other template sets may reuse roles for many templates. Therefore it will be necessary to refer to particular templates in mapping descriptions, not only to roles. As I can see now, such reference is not included in your examples.
In the example provided the template is referenced in the TemplateMap. This then has a list of RoleMappings rather than each role mapping duplicating the template reference.
If I'm not mistaken, your example so far doesn't cover this third possible part? Is it possible to express such correspondence in RDF/OWL?
I don't understand your statement 'not explicitly instantiated'. If this is the discussion on optional roles then this is true, there is no support for optional roles as that has different semantics and is not a single expansion. If this is just two template roles with the same class reference then each one can be defined independently. If this is two template roles mapped to a single TIP parameter then this is covered.

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Structures for the publication of TIPs

#14 Post by vvagr »

Sorry, I've overlooked tips:hasTemplate predicate :-(
If this is just two template roles with the same class reference then each one can be defined independently.
Yes, I'm thinking about this case. The same class reference is OK, but it is essential part of the pattern semantics that these roles are occupied by the same member of the class referred to in mapping. Of course it can be left to expansion software to generate "dummy" URI for role population and add necessary classification.

But probable there can be a case when in a complex pattern the class reference for two pairs of roles is the same, but business meaning requires two distinct role occupiers for each pair? Theoretically it is possible.
there is no support for optional roles as that has different semantics and is not a single expansion
As well as I understand, it will be really simple to extend your suggestions to cover alternative mappings and thus allow for optional roles. tips:templateMappings can include several named tips:templateMapping elements, each one consisting of several tips:templateMap.

Andrew.Prosser

Re: Structures for the publication of TIPs

#15 Post by Andrew.Prosser »

Are you trying to manage constraints on parameters for the TIP? if two roles in the template expansion must have the same value then surely there is one parameter for the TIP?

I'm not sure alternative mappings can be used for optional roles given that each mapping would have different meaning and surely a TIP is supposed to give consistent meaning??

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Structures for the publication of TIPs

#16 Post by vvagr »

Are you trying to manage constraints on parameters for the TIP? if two roles in the template expansion must have the same value then surely there is one parameter for the TIP?
No, here I'm thinking about template roles which are not mapped to TIP parameters, like the roles under the existential quantifier in template axiom.

As well as I understand, in a Coating Color TIP two templates have a member of EXTERNAL COATING class occupying a role in each of the templates, but these roles are not mapped to TIP parameters.
I'm not sure alternative mappings can be used for optional roles given that each mapping would have different meaning and surely a TIP is supposed to give consistent meaning??
As we've discussed, single mapping may be an additional restriction for TIPs, but format for broader TSPs should be the same, and TSPs so far are designed with possibility for optional parameters.

Multiple mappings are a way to allow optional parameters, but they have other uses too! I think that even for TIPs multiple mappings can become useful if two or more template sets continue their coexistence for some time.

Andrew.Prosser

Re: Structures for the publication of TIPs

#17 Post by Andrew.Prosser »

Victor....

I don't really understand what you are asking in your post. Do you have an example? Either the TIP has a single role that is mapped to multiple templates or the roles have fixed values.
Yes, I'm thinking about this case. The same class reference is OK, but it is essential part of the pattern semantics that these roles are occupied by the same member of the class referred to in mapping. Of course it can be left to expansion software to generate "dummy" URI for role population and add necessary classification.
either there is a single TIP parameter constrained as an EXTERNAL COATING mapped to roles of 2 templates or no TIP parameters and both templates have a fixed class mapped to their roles. :?:

Andy

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Structures for the publication of TIPs

#18 Post by vvagr »

Andy,

I'm looking at templates comprising Coating Color TIP in a TIP database. There are templates AssemblyOfIndividual and ClassificationOfIndividual.

As well as I can understand it, both templates have a fixed individual mapped to their roles and restricted by an EXTERNAL COATING class. And there are no third TIP parameter for this individual.

Am I wrong here?

Andrew.Prosser

Re: Structures for the publication of TIPs

#19 Post by Andrew.Prosser »

My understanding from the TIP is that 3 parameters are required...

TIP :

Param 1: Individual (with coating)
Param 2 : Individual (coating) - constrained/declared to be a of type 'EXTERNAL COATING'
Param 3 : Classification for 'Individual (coating)' - constrained/declared to be a subclass of 'EXTERNAL COATING CLASS'

The TIP parameters must restrict the data beyond the restrictions placed on the original template roles. I was surprised to see coatings modelled as individuals since this then enforces identity/lifecycle maintenance throughout :(

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Structures for the publication of TIPs

#20 Post by vvagr »

This parameter will resolve the problem, but I can not see it in the model. More then that, such third role is completely unnecessary. In templates existential quantifier (or Scolem variable) is used to hide exactly this type of unnecessary data items. Why do we return them into patterns?

Post Reply