Object Model / Ontology Management - how we kick it off.
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
1. It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc). 2. It should be dynamic (i.e. capable of evolving through time). 3. It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced. 4. It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated. 5. It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
*Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?*
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
Roger,
One slight extension/modification I would like to see is that I would really like to see the polymorphism more like Java interface implementation than Java class extension (i.e. "multiple inheritance" may be a good thing if suitably controlled, e.g. by namespaces). My point here is that I want it to be easy for our data providers to make use of all relevant polymorphisms (extensions) when serving their data. Darwin Core is my model here. If we develop a range of extension vocabularies to augment Darwin Core in describing a taxon occurrence, providers should be able to serve data including any subset of those vocabularies.
This may be so obvious as not to need saying, but I wanted to be sure it was captured.
By the way, I thoroughly agree with your other points and believe that we should focus on the small shared vocabulary you describe. If we define these anchor points, subgroups can address everything that we need to flesh out these classes for use in applications (including defining properties that relate their objects to objects of other classes).
Thanks,
Donald
--------------------------------------------------------------- Donald Hobern (dhobern@gbif.org) Programme Officer for Data Access and Database Interoperability Global Biodiversity Information Facility Secretariat Universitetsparken 15, DK-2100 Copenhagen, Denmark Tel: +45-35321483 Mobile: +45-28751483 Fax: +45-35321480 ---------------------------------------------------------------
_____
From: Tdwg-tag-bounces@lists.tdwg.org [mailto:Tdwg-tag-bounces@lists.tdwg.org] On Behalf Of Roger Hyam Sent: 22 February 2006 16:50 To: Tdwg-tag@lists.tdwg.org Subject: [Tdwg-tag] Object Model / Ontology Management - how we kick it off.
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
1. It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc). 2. It should be dynamic (i.e. capable of evolving through time). 3. It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced. 4. It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated. 5. It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
I agree with Donald about multiple inheritance, and with you concerning the simple prescription of anchor points. These in turn can act as a models for the subgroups to elaborate.
On 2/22/06, Donald Hobern dhobern@gbif.org wrote:
Roger,
One slight extension/modification I would like to see is that I would really like to see the polymorphism more like Java interface implementation than Java class extension (i.e. "multiple inheritance" may be a good thing if suitably controlled, e.g. by namespaces). My point here is that I want it to be easy for our data providers to make use of all relevant polymorphisms (extensions) when serving their data. Darwin Core is my model here. If we develop a range of extension vocabularies to augment Darwin Core in describing a taxon occurrence, providers should be able to serve data including any subset of those vocabularies.
This may be so obvious as not to need saying, but I wanted to be sure it was captured.
By the way, I thoroughly agree with your other points and believe that we should focus on the small shared vocabulary you describe. If we define these anchor points, subgroups can address everything that we need to flesh out these classes for use in applications (including defining properties that relate their objects to objects of other classes).
Thanks,
Donald
Donald Hobern (dhobern@gbif.org) Programme Officer for Data Access and Database Interoperability Global Biodiversity Information Facility Secretariat Universitetsparken 15, DK-2100 Copenhagen, Denmark Tel: +45-35321483 Mobile: +45-28751483 Fax: +45-35321480
*From:* Tdwg-tag-bounces@lists.tdwg.org [mailto: Tdwg-tag-bounces@lists.tdwg.org] *On Behalf Of *Roger Hyam *Sent:* 22 February 2006 16:50 *To:* Tdwg-tag@lists.tdwg.org *Subject:* [Tdwg-tag] Object Model / Ontology Management - how we kick it off.
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
- It should be representation independent (i.e. we should be able
to move it between 'languages' UML, OWL, BNF etc). 2. It should be dynamic (i.e. capable of evolving through time). 3. It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced. 4. It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated. 5. It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
*Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?*
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
--
Roger Hyam
Technical Architect
Taxonomic Databases Working Group
roger@tdwg.org
+44 1578 722782
Tdwg-tag mailing list Tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag_lists.tdwg.org
Thanks for that Donald. I am glad you agree on the core vocabulary.
The use of the term 'vocabulary' is where I think we can try and address the other point you raise regarding interfaces vs extensions. It leads neatly into my next question to the list that I'll pop in another thread.
Donald Hobern wrote:
Roger,
One slight extension/modification I would like to see is that I would really like to see the polymorphism more like Java interface implementation than Java class extension (i.e. "multiple inheritance" may be a good thing if suitably controlled, e.g. by namespaces). My point here is that I want it to be easy for our data providers to make use of all relevant polymorphisms (extensions) when serving their data. Darwin Core is my model here. If we develop a range of extension vocabularies to augment Darwin Core in describing a taxon occurrence, providers should be able to serve data including any subset of those vocabularies.
This may be so obvious as not to need saying, but I wanted to be sure it was captured.
By the way, I thoroughly agree with your other points and believe that we should focus on the small shared vocabulary you describe. If we define these anchor points, subgroups can address everything that we need to flesh out these classes for use in applications (including defining properties that relate their objects to objects of other classes).
Thanks,
Donald
Donald Hobern (dhobern@gbif.org mailto:dhobern@gbif.org) Programme Officer for Data Access and Database Interoperability Global Biodiversity Information Facility Secretariat Universitetsparken 15, DK-2100 Copenhagen, Denmark Tel: +45-35321483 Mobile: +45-28751483 Fax: +45-35321480
*From:* Tdwg-tag-bounces@lists.tdwg.org [mailto:Tdwg-tag-bounces@lists.tdwg.org] *On Behalf Of *Roger Hyam *Sent:* 22 February 2006 16:50 *To:* Tdwg-tag@lists.tdwg.org *Subject:* [Tdwg-tag] Object Model / Ontology Management - how we kick it off.
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
- It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc).
- It should be dynamic (i.e. capable of evolving through time).
- It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced.
- It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated.
- It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
*Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?*
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
--
Roger Hyam Technical Architect Taxonomic Databases Working Group
http://www.tdwg.org roger@tdwg.org mailto:roger@tdwg.org
+44 1578 722782
Roger,
I think I agree with most of your points (also from previous messages).
Concerning the representation independent object model, I would suggest the same approach taken by CIDOC CRM:
http://cidoc.ics.forth.gr/docs/cidoc_crm_version_4.2.pdf
It's not easy to find such clear and well documented modelling work.
Although at a first glance it could fall into the "ontology-at-the- level-of-laws-of-physics" category, I won't make that judgment because CIDOC's scope is definitely broader than ours. Anyway, what I'm suggesting is to use the same approach and the same kind of documentation. Using and extending CIDOC is a completely different thing - probably interesting (I think), but something that could even be evaluated and addressed at another stage.
Regards, -- Renato
On 22 Feb 2006 at 15:50, Roger Hyam wrote:
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
- It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc).
- It should be dynamic (i.e. capable of evolving through time).
- It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced.
- It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated.
- It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
Hi Renato,
Thanks for that contribution. I have had CIDOC on my list since the St Petersburg meeting last year. We definitely need to address how we make use of it or integrate with it. My worry is looking for implementations that use it. I am not aware of organisations sharing data on the basis of schemas derived from the CIDOC model - this is worrying as the ontology has been under development for 10 years - it may also be that I am ignorant.
Does anyone on the list have practical experience of using CIDOC?
Thanks,
Roger
Renato De Giovanni wrote:
Roger,
I think I agree with most of your points (also from previous messages).
Concerning the representation independent object model, I would suggest the same approach taken by CIDOC CRM:
http://cidoc.ics.forth.gr/docs/cidoc_crm_version_4.2.pdf
It's not easy to find such clear and well documented modelling work.
Although at a first glance it could fall into the "ontology-at-the- level-of-laws-of-physics" category, I won't make that judgment because CIDOC's scope is definitely broader than ours. Anyway, what I'm suggesting is to use the same approach and the same kind of documentation. Using and extending CIDOC is a completely different thing - probably interesting (I think), but something that could even be evaluated and addressed at another stage.
Regards,
Renato
On 22 Feb 2006 at 15:50, Roger Hyam wrote:
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
- It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc).
- It should be dynamic (i.e. capable of evolving through time).
- It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced.
- It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated.
- It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
Tdwg-tag mailing list Tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag_lists.tdwg.org
Continuing on from my last message the way the CIDOC modeling good and is presented is not unlike the way the DublinCore work is carried out - and the DC terms are also specified in RDF using a PURL!
http://www.dublincore.org/documents/dcmi-terms/
Renato De Giovanni wrote:
Roger,
I think I agree with most of your points (also from previous messages).
Concerning the representation independent object model, I would suggest the same approach taken by CIDOC CRM:
http://cidoc.ics.forth.gr/docs/cidoc_crm_version_4.2.pdf
It's not easy to find such clear and well documented modelling work.
Although at a first glance it could fall into the "ontology-at-the- level-of-laws-of-physics" category, I won't make that judgment because CIDOC's scope is definitely broader than ours. Anyway, what I'm suggesting is to use the same approach and the same kind of documentation. Using and extending CIDOC is a completely different thing - probably interesting (I think), but something that could even be evaluated and addressed at another stage.
Regards,
Renato
On 22 Feb 2006 at 15:50, Roger Hyam wrote:
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
- It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc).
- It should be dynamic (i.e. capable of evolving through time).
- It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced.
- It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated.
- It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
Tdwg-tag mailing list Tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag_lists.tdwg.org
Hi all,
I think I also agree with what you are saying, but I suppose is also too general and that in the details can arise differences. Specially if we want TAG only to define a shared vocabulary or classes, how complicate should these classes be?
Also something I would like to have as a kind of requirement is the possible automatic derivation of XML schemas from whatever modeling language we choose now.
In the real world there will be lot of times where we will need to create XML schemas from these TAG classes and the TDWG subgroups classes. The best example, as pointed out by Flip in a previous message, is that we will need to implement our models in GML. Why? Because all OGC standards are based on it so if we want to play with them we will have to use their rules. I think he also pointed out about a document to do GML app schemas from UML models with automatic derivation. There is also an article from IBM about modeling XML using UML (http://www-128.ibm.com/ developerworks/xml/library/x-wxxm23/). In this article is discussed the need to do automatic derivation to keep synchronize the modeling task with the programming task.
I think this is importance so maybe it is worth to include it as a kind of requirement. But again, depending on how far TAG wants to go this might be more a discussion for the different TDWG subgroups.
Javier.
PD: Here are the links: http://www-128.ibm.com/developerworks/xml/library/x-wxxm23/ http://www-128.ibm.com/developerworks/xml/library/x-wxxm24/ http://www-128.ibm.com/developerworks/xml/library/x-wxxm25/ http://www-128.ibm.com/developerworks/xml/library/x-wxxm26/
GML From UML https://www.seegrid.csiro.au/twiki/bin/view/Xmml/UmlGml
On 22/02/2006, at 16:50, Roger Hyam wrote:
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong). It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc). It should be dynamic (i.e. capable of evolving through time). It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced. It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated. It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it. My first Question is:
Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
--
Roger Hyam Technical Architect Taxonomic Databases Working Group
http://www.tdwg.org roger@tdwg.org
+44 1578 722782
Tdwg-tag mailing list Tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag_lists.tdwg.org
Hi Javier,
Sorry for the delay in responding to this one.
Javier de la Torre wrote:
I think I also agree with what you are saying, but I suppose is also too general and that in the details can arise differences. Specially if we want TAG only to define a shared vocabulary or classes, how complicate should these classes be?
This is part of my plan. It is easier to agree on the general things and so I want to fix these things in stone and slowly move towards the detailed and more controversial subjects. We may find that some of the controversial subjects can be avoided all together or, when we have a framework of agreement in place, it may be possible to have a polymorphic approach to some well defined areas.
This is counter intuitive for most of us (including me) as we tend to be analysts who look for the problems and exceptions in every case - and are therefore unlikely to ever reach agreement.
Your suggestions about auto generation of XML from UML are great. I have added them to a wiki page here:
http://www.tdwg.hyam.net/twiki/bin/view/TAG/AutoGenerationOfXMLSchema
and added that to a list of things to be discussed here:
http://www.tdwg.hyam.net/twiki/bin/view/TAG/TagDiscussionRoadMap
All the best,
Roger
On 22/02/2006, at 16:50, Roger Hyam wrote:
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
- It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc).
- It should be dynamic (i.e. capable of evolving through time).
- It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced.
- It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated.
- It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
*Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?*
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
--
Roger Hyam Technical Architect Taxonomic Databases Working Group
http://www.tdwg.org roger@tdwg.org
+44 1578 722782
Tdwg-tag mailing list Tdwg-tag@lists.tdwg.org mailto:Tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag_lists.tdwg.org
Roger,
I'm going to stick my neck on the block here.
GBIF clearly has an urgent need for a vocabulary for classes of data accessed and served by the Data Portal and for a vocabulary of the concepts or properties which we handle in connection with each of these classes. The GBIF Data Portal Strategy document includes the idea of a Schema Repository, which was originally conceived as a way to manage the various XML schema definitions of interest to us, but everything is moving so fast that it now seems much more plausible than it did before for us to use a basic ontology of biodiversity data classes and the properties which we need to manage in relation to them (including all of the different syntactic representations of semantically related properties).
I would like to see this work take place under the control of TDWG, but in the mean time I need to make some decisions which can guide the development of GBIF's portal and services over the next few months, so I have taken a pass at representing the simplest core ontology I can for the data portal. There are certainly several obvious areas for discussion, but much of it simply reflects what I believe is implicit in the current TDWG data standards.
I have also put together a sketchy proposal for how GBIF might use such a core ontology as the basis for the function which was originally suggested for the Schema Repository. I would value comments on any or all of these things - allowing for the fact that the current description is little more than a stream-of-consciousness first pass.
You can see the materials at the Data Portal wiki at: http://wiki.gbif.org/dadiwiki/wikka.php?wakka=SchemaRepository
Follow the links for the CoreOntology, PropertyStore, SchemaRepositoryFacade and SchemaRepositoryUseCases to get some idea of what I am thinking.
Thanks,
Donald
--------------------------------------------------------------- Donald Hobern (dhobern@gbif.org) Programme Officer for Data Access and Database Interoperability Global Biodiversity Information Facility Secretariat Universitetsparken 15, DK-2100 Copenhagen, Denmark Tel: +45-35321483 Mobile: +45-28751483 Fax: +45-35321480 ---------------------------------------------------------------
_____
From: Tdwg-tag-bounces@lists.tdwg.org [mailto:Tdwg-tag-bounces@lists.tdwg.org] On Behalf Of Roger Hyam Sent: 01 March 2006 13:35 To: Javier de la Torre Cc: Tdwg-tag@lists.tdwg.org Subject: Re: [Tdwg-tag] Automatic derivation of XML schemas from models
Hi Javier,
Sorry for the delay in responding to this one.
Javier de la Torre wrote:
I think I also agree with what you are saying, but I suppose is also too general and that in the details can arise differences. Specially if we want TAG only to define a shared vocabulary or classes, how complicate should these classes be?
This is part of my plan. It is easier to agree on the general things and so I want to fix these things in stone and slowly move towards the detailed and more controversial subjects. We may find that some of the controversial subjects can be avoided all together or, when we have a framework of agreement in place, it may be possible to have a polymorphic approach to some well defined areas.
This is counter intuitive for most of us (including me) as we tend to be analysts who look for the problems and exceptions in every case - and are therefore unlikely to ever reach agreement.
Your suggestions about auto generation of XML from UML are great. I have added them to a wiki page here:
http://www.tdwg.hyam.net/twiki/bin/view/TAG/AutoGenerationOfXMLSchema
and added that to a list of things to be discussed here:
http://www.tdwg.hyam.net/twiki/bin/view/TAG/TagDiscussionRoadMap
All the best,
Roger
On 22/02/2006, at 16:50, Roger Hyam wrote:
Hi All,
It is generally agreed that we need an representation independent object model or ontology of some kind. I would like to put together a list of the things that need to be agreed or investigated in order to do this.
Firstly the things I believe we can all agree on (stop me if I am wrong).
1. It should be representation independent (i.e. we should be able to move it between 'languages' UML, OWL, BNF etc). 2. It should be dynamic (i.e. capable of evolving through time). 3. It should be polymorphic. This is a result of it being dynamic. There will, at a minimum, be multiple version of any one part of the model when new version are introduced. 4. It should NOT attempt to be omniscient i.e. it will not cover everything in our domain, only the parts that need to be communicated. 5. It will be managed in a distributed fashion. Different teams will take responsibility for different parts of it.
My first Question is:
Does the centralization of the ontology need to go beyond a small shared vocabulary of terms or base classes?
I envisage this ontology containing things like Collection, Specimen, TaxonConcept, TaxonName but not defining the detailed structure of these objects. It would contain a maximum of a few 10's of objects and properties. TDWG subgroups would be responsible for building ontologies that extend these base objects but that generally didn't refer to each other - only to the core. If this is true then I think the definition of the top level object falls within the remit of the TAG ( in consultation with others).
If this is not a valid way forward what are the alternatives?
Are their questions we should ask before this one?
Once again I'd be grateful for your thoughts.
Roger
participants (5)
-
Donald Hobern -
Javier de la Torre -
John R. WIECZOREK -
Renato De Giovanni -
Roger Hyam