Manchester Syntax unsuited?

[Bud P. Bruegger]

Hello,

I found this archived message by Timothy Redmond:

> Something that I forgot to mention.  The Manchester OWL syntax is very
> good - it is used throughout Protege.  But as a format for saving
> ontologies for real work, it is not recommended.  The problem is that
> there are a couple of constructs that cannot be represented in the
> Manchester OWL syntax.  The ones that I know about are annotations of
> entities that are not declared and general concept inclusion axioms.
> These axioms will get dropped when you save to this format.  The
> latest Protege gives a warning when select the Manchester OWL syntax
> for saving ontologies.

I'd like to use some follow-up questions:

* are there other limitations than the ones listed? (as implied by "The
  ones that I know about")

* how serious are these limitations in real life?  For example, at
  least naively, I thought that I could live without annotations of
  undeclared entities (I may not use annotations at all), but I'm not
  sure what I would loose when dropping general concept inclusions from
  my ontology.

* Are these limitations inherent in the Manchester Syntax or is it
  likely that there will be a future version that can represent
  these constructs?

many thanks for your advice!

-b
_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Timothy Redmond]

> I'd like to use some follow-up questions:
>
> * are there other limitations than the ones listed? (as implied by "The
>    ones that I know about")

I don't know.
> * how serious are these limitations in real life?  For example, at
>    least naively, I thought that I could live without annotations of
>    undeclared entities (I may not use annotations at all), but I'm not
>    sure what I would loose when dropping general concept inclusions from
>    my ontology.

Probably in real life these limitations are not that severe.  
Annotations are very important but generally the entity declaration is
somewhere nearby.  GCI's are not that common and you will know when you
are using them.  But the Protege team in general takes data corruption
issues very seriously.  The idea that you could make some changes to an
ontology and lose them when you save it is bad.

Matthew indicated that at some time in the future there might be an OWL
api function that would be able to check if a particular serialization
format would lose data for a particular ontology.  If this capability
was implemented then we could change the warning for the Manchester OWL
syntax to only come up if data would be lost.

> * Are these limitations inherent in the Manchester Syntax or is it
>    likely that there will be a future version that can represent
>    these constructs?


These limitations are part of the specification of the Manchester syntax
so they are unlikely to go away.

-Timothy


On 06/21/2011 11:16 AM, Bud P. Bruegger wrote:

> Hello,
>
> I found this archived message by Timothy Redmond:
>
>> Something that I forgot to mention.  The Manchester OWL syntax is very
>> good - it is used throughout Protege.  But as a format for saving
>> ontologies for real work, it is not recommended.  The problem is that
>> there are a couple of constructs that cannot be represented in the
>> Manchester OWL syntax.  The ones that I know about are annotations of
>> entities that are not declared and general concept inclusion axioms.
>> These axioms will get dropped when you save to this format.  The
>> latest Protege gives a warning when select the Manchester OWL syntax
>> for saving ontologies.
> I'd like to use some follow-up questions:
>
> * are there other limitations than the ones listed? (as implied by "The
>    ones that I know about")
>
> * how serious are these limitations in real life?  For example, at
>    least naively, I thought that I could live without annotations of
>    undeclared entities (I may not use annotations at all), but I'm not
>    sure what I would loose when dropping general concept inclusions from
>    my ontology.
>
> * Are these limitations inherent in the Manchester Syntax or is it
>    likely that there will be a future version that can represent
>    these constructs?
>
> many thanks for your advice!
>
> -b
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Martin O'Connor]

The Manchester Syntax is an excellent OWL presentation syntax and works very well in GUI tools and in text publications
for displaying individual OWL axioms.

In general, the choice of a particular OWL serialization should not be driven by readability since tools like Protege
are used to display and edit the ontology in a user friendly way.

OWL ontologies are built from an unsorted bag of axioms and it is often not that easy to make sense of an ontology by
isolated inspection of axioms, no matter how readable the individual axioms may be. Protege does all the hard work of
reading this bag of axioms and assembling an overall view of the axioms in an ontology and their relationships. In other
words, having a human readable serialization of your entire ontology may not be all that advantageous.

However, you may have a particular use case driving your decision.

Martin

On 6/21/2011 11:50 AM, Timothy Redmond wrote:

>
>> I'd like to use some follow-up questions:
>>
>> * are there other limitations than the ones listed? (as implied by "The
>>    ones that I know about")
>
> I don't know.
>> * how serious are these limitations in real life?  For example, at
>>    least naively, I thought that I could live without annotations of
>>    undeclared entities (I may not use annotations at all), but I'm not
>>    sure what I would loose when dropping general concept inclusions from
>>    my ontology.
>
> Probably in real life these limitations are not that severe.  Annotations are very important but generally the entity
> declaration is somewhere nearby.  GCI's are not that common and you will know when you are using them.  But the
> Protege team in general takes data corruption issues very seriously.  The idea that you could make some changes to an
> ontology and lose them when you save it is bad.
>
> Matthew indicated that at some time in the future there might be an OWL api function that would be able to check if a
> particular serialization format would lose data for a particular ontology.  If this capability was implemented then we
> could change the warning for the Manchester OWL syntax to only come up if data would be lost.
>
>> * Are these limitations inherent in the Manchester Syntax or is it
>>    likely that there will be a future version that can represent
>>    these constructs?
>
>
> These limitations are part of the specification of the Manchester syntax so they are unlikely to go away.
>
> -Timothy
>
>
> On 06/21/2011 11:16 AM, Bud P. Bruegger wrote:
>> Hello,
>>
>> I found this archived message by Timothy Redmond:
>>
>>> Something that I forgot to mention.  The Manchester OWL syntax is very
>>> good - it is used throughout Protege.  But as a format for saving
>>> ontologies for real work, it is not recommended.  The problem is that
>>> there are a couple of constructs that cannot be represented in the
>>> Manchester OWL syntax.  The ones that I know about are annotations of
>>> entities that are not declared and general concept inclusion axioms.
>>> These axioms will get dropped when you save to this format.  The
>>> latest Protege gives a warning when select the Manchester OWL syntax
>>> for saving ontologies.
>> I'd like to use some follow-up questions:
>>
>> * are there other limitations than the ones listed? (as implied by "The
>>    ones that I know about")
>>
>> * how serious are these limitations in real life?  For example, at
>>    least naively, I thought that I could live without annotations of
>>    undeclared entities (I may not use annotations at all), but I'm not
>>    sure what I would loose when dropping general concept inclusions from
>>    my ontology.
>>
>> * Are these limitations inherent in the Manchester Syntax or is it
>>    likely that there will be a future version that can represent
>>    these constructs?
>>
>> many thanks for your advice!
>>
>> -b
>> _______________________________________________
>> protege-owl mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>
>> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03
>
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03
>

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Alan Ruttenberg-2]

On Tue, Jun 21, 2011 at 3:50 PM, Martin O'Connor
<[hidden email]> wrote:
> The Manchester Syntax is an excellent OWL presentation syntax and works very
> well in GUI tools and in text publications for displaying individual OWL
> axioms.

Just to clarify a certain matter, note that protege only *sometimes*
uses manchester syntax. Manchester syntax as in the spec does not
sanction the use of labels in place of entity URIs, which protege
(thankfully) allows. So what you see when you serialize to Manchester
syntax may not be nearly as readably as what you see in protege, since
the URIs will be used throughout.

-Alan

>
> In general, the choice of a particular OWL serialization should not be
> driven by readability since tools like Protege are used to display and edit
> the ontology in a user friendly way.
>
> OWL ontologies are built from an unsorted bag of axioms and it is often not
> that easy to make sense of an ontology by isolated inspection of axioms, no
> matter how readable the individual axioms may be. Protege does all the hard
> work of reading this bag of axioms and assembling an overall view of the
> axioms in an ontology and their relationships. In other words, having a
> human readable serialization of your entire ontology may not be all that
> advantageous.
>
> However, you may have a particular use case driving your decision.
>
> Martin
>
> On 6/21/2011 11:50 AM, Timothy Redmond wrote:
>>
>>> I'd like to use some follow-up questions:
>>>
>>> * are there other limitations than the ones listed? (as implied by "The
>>>   ones that I know about")
>>
>> I don't know.
>>>
>>> * how serious are these limitations in real life?  For example, at
>>>   least naively, I thought that I could live without annotations of
>>>   undeclared entities (I may not use annotations at all), but I'm not
>>>   sure what I would loose when dropping general concept inclusions from
>>>   my ontology.
>>
>> Probably in real life these limitations are not that severe.  Annotations
>> are very important but generally the entity declaration is somewhere nearby.
>>  GCI's are not that common and you will know when you are using them.  But
>> the Protege team in general takes data corruption issues very seriously.
>>  The idea that you could make some changes to an ontology and lose them when
>> you save it is bad.
>>
>> Matthew indicated that at some time in the future there might be an OWL
>> api function that would be able to check if a particular serialization
>> format would lose data for a particular ontology.  If this capability was
>> implemented then we could change the warning for the Manchester OWL syntax
>> to only come up if data would be lost.
>>
>>> * Are these limitations inherent in the Manchester Syntax or is it
>>>   likely that there will be a future version that can represent
>>>   these constructs?
>>
>>
>> These limitations are part of the specification of the Manchester syntax
>> so they are unlikely to go away.
>>
>> -Timothy
>>
>>
>> On 06/21/2011 11:16 AM, Bud P. Bruegger wrote:
>>>
>>> Hello,
>>>
>>> I found this archived message by Timothy Redmond:
>>>
>>>> Something that I forgot to mention.  The Manchester OWL syntax is very
>>>> good - it is used throughout Protege.  But as a format for saving
>>>> ontologies for real work, it is not recommended.  The problem is that
>>>> there are a couple of constructs that cannot be represented in the
>>>> Manchester OWL syntax.  The ones that I know about are annotations of
>>>> entities that are not declared and general concept inclusion axioms.
>>>> These axioms will get dropped when you save to this format.  The
>>>> latest Protege gives a warning when select the Manchester OWL syntax
>>>> for saving ontologies.
>>>
>>> I'd like to use some follow-up questions:
>>>
>>> * are there other limitations than the ones listed? (as implied by "The
>>>   ones that I know about")
>>>
>>> * how serious are these limitations in real life?  For example, at
>>>   least naively, I thought that I could live without annotations of
>>>   undeclared entities (I may not use annotations at all), but I'm not
>>>   sure what I would loose when dropping general concept inclusions from
>>>   my ontology.
>>>
>>> * Are these limitations inherent in the Manchester Syntax or is it
>>>   likely that there will be a future version that can represent
>>>   these constructs?
>>>
>>> many thanks for your advice!
>>>
>>> -b
>>> _______________________________________________
>>> protege-owl mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>>
>>> Instructions for unsubscribing:
>>> http://protege.stanford.edu/doc/faq.html#01a.03
>>
>> _______________________________________________
>> protege-owl mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>
>> Instructions for unsubscribing:
>> http://protege.stanford.edu/doc/faq.html#01a.03
>>
>
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03
>

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Bud P. Bruegger]

Thanks a lot, Timothy.  One more followup below

On Tue, Jun 21, 2011 at 8:50 PM, Timothy Redmond <[hidden email]> wrote:

>> * Are these limitations inherent in the Manchester Syntax or is it
>>   likely that there will be a future version that can represent
>>   these constructs?
>
> These limitations are part of the specification of the Manchester syntax so
> they are unlikely to go away.

But could it be fixed by a new version of the spec that, for example,
introduces another keyword or similar?

I'm somehow surprised that Manchester Syntax isn't complete and was
wondering why.  Was it an oversight, was it because it would be
incompatible with basic design decisions?   In the former case, a new
version of the spec could fix the problem, in the latter, there is no
hope..

tnx
-b
_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Bud P. Bruegger]

> So what you see when you serialize to Manchester
> syntax may not be nearly as readably as what you see in protege, since
> the URIs will be used throughout.

Hi Alan,

it seems possible to get something readable when manually setting the
right prefix in protege:

with Prefix: : <http://www.eid-ontology.org/ontologies/citizens/> set,
an excerpt of some of my experiments looks like this (serialized with
Protege).


Class: Citizen

    SubClassOf:
        Person


-b
_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Timothy Redmond]

> I'm somehow surprised that Manchester Syntax isn't complete and was
> wondering why.  Was it an oversight, was it because it would be
> incompatible with basic design decisions?   In the former case, a new
> version of the spec could fix the problem, in the latter, there is no
> hope..

My guess is that there is no hope.

I was surprised to find that it wasn't complete also.  But once this is
identified, I don't see it as that big a deal.  I think that Martin's
post about this was very good.  The serialization format for an OWL
ontology simply isn't that important.  The main advantage of the
Manchester OWL syntax is for presentation in e-mails and journal articles.

If I was going to go back to editing OWL ontologies with a text editor
(as I did before I knew Protege) I would probably choose either the
functional OWL syntax or the OWL/XML syntax.  But I think that there are
now better tools for editing OWL ontologies than a text editor.

-Timothy



On 06/22/2011 04:05 AM, Bud P. Bruegger wrote:

> Thanks a lot, Timothy.  One more followup below
>
> On Tue, Jun 21, 2011 at 8:50 PM, Timothy Redmond<[hidden email]>  wrote:
>
>>> * Are these limitations inherent in the Manchester Syntax or is it
>>>    likely that there will be a future version that can represent
>>>    these constructs?
>> These limitations are part of the specification of the Manchester syntax so
>> they are unlikely to go away.
> But could it be fixed by a new version of the spec that, for example,
> introduces another keyword or similar?
>
> I'm somehow surprised that Manchester Syntax isn't complete and was
> wondering why.  Was it an oversight, was it because it would be
> incompatible with basic design decisions?   In the former case, a new
> version of the spec could fix the problem, in the latter, there is no
> hope..
>
> tnx
> -b
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Alan Ruttenberg-2]

On Wed, Jun 22, 2011 at 8:55 AM, Bud P. Bruegger
<[hidden email]> wrote:

>> So what you see when you serialize to Manchester
>> syntax may not be nearly as readably as what you see in protege, since
>> the URIs will be used throughout.
>
> Hi Alan,
>
> it seems possible to get something readable when manually setting the
> right prefix in protege:
>
> with Prefix: : <http://www.eid-ontology.org/ontologies/citizens/> set,
> an excerpt of some of my experiments looks like this (serialized with
> Protege).
>
>
> Class: Citizen
>
>    SubClassOf:
>        Person
>

Yes. However if your URIs are as OBOs are, namely things like
OBI_00011007, and you use label display in protege then when you
serialize you will see the OBI_00011007 rather than the label you see
in protege.
-Alan

>
> -b
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03
>
_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Alan Ruttenberg-2]

On Tue, Jun 21, 2011 at 2:50 PM, Timothy Redmond <[hidden email]> wrote:

>
>> I'd like to use some follow-up questions:
>>
>> * are there other limitations than the ones listed? (as implied by "The
>>   ones that I know about")
>
> I don't know.
>>
>> * how serious are these limitations in real life?  For example, at
>>   least naively, I thought that I could live without annotations of
>>   undeclared entities (I may not use annotations at all), but I'm not
>>   sure what I would loose when dropping general concept inclusions from
>>   my ontology.
>
> Probably in real life these limitations are not that severe.  Annotations
> are very important but generally the entity declaration is somewhere nearby.
>  GCI's are not that common and you will know when you are using them.  But
> the Protege team in general takes data corruption issues very seriously.
>  The idea that you could make some changes to an ontology and lose them when
> you save it is bad.
>
> Matthew indicated that at some time in the future there might be an OWL api
> function that would be able to check if a particular serialization format
> would lose data for a particular ontology.  If this capability was
> implemented then we could change the warning for the Manchester OWL syntax
> to only come up if data would be lost.
>
>> * Are these limitations inherent in the Manchester Syntax or is it
>>   likely that there will be a future version that can represent
>>   these constructs?
>
>
> These limitations are part of the specification of the Manchester syntax so
> they are unlikely to go away.

The GCI restriction can be worked around, and probably should be. The
idea is to create a named class for one of the sides of the GCI and
then make an assertion on that class. Protege could manage the round
trip by adding an annotation to the class.

I hadn't realized about the other case :(. Since I actually care about
using URIs that are not entities I'll think about whether there is
some workaround for that.

-Alan

>
> -Timothy
>
>
> On 06/21/2011 11:16 AM, Bud P. Bruegger wrote:
>>
>> Hello,
>>
>> I found this archived message by Timothy Redmond:
>>
>>> Something that I forgot to mention.  The Manchester OWL syntax is very
>>> good - it is used throughout Protege.  But as a format for saving
>>> ontologies for real work, it is not recommended.  The problem is that
>>> there are a couple of constructs that cannot be represented in the
>>> Manchester OWL syntax.  The ones that I know about are annotations of
>>> entities that are not declared and general concept inclusion axioms.
>>> These axioms will get dropped when you save to this format.  The
>>> latest Protege gives a warning when select the Manchester OWL syntax
>>> for saving ontologies.
>>
>> I'd like to use some follow-up questions:
>>
>> * are there other limitations than the ones listed? (as implied by "The
>>   ones that I know about")
>>
>> * how serious are these limitations in real life?  For example, at
>>   least naively, I thought that I could live without annotations of
>>   undeclared entities (I may not use annotations at all), but I'm not
>>   sure what I would loose when dropping general concept inclusions from
>>   my ontology.
>>
>> * Are these limitations inherent in the Manchester Syntax or is it
>>   likely that there will be a future version that can represent
>>   these constructs?
>>
>> many thanks for your advice!
>>
>> -b
>> _______________________________________________
>> protege-owl mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>
>> Instructions for unsubscribing:
>> http://protege.stanford.edu/doc/faq.html#01a.03
>
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03
>

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

[Thomas Russ]

On Jun 21, 2011, at 11:50 AM, Timothy Redmond wrote:

>>
>> * are there other limitations than the ones listed? (as implied by "The
>>   ones that I know about")
>
> I don't know.

Open world semantics strike again!

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03