You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jena.apache.org by "Scoggan, Tom" <to...@baml.com> on 2016/04/28 16:56:08 UTC
Namespace prefixes in Jena Models
When writing a Model to an RDF string, I currently do something like this:
1. Use the model.listNameSpaces() method to get the list of namespaces used --- this doesn't include namespaces used in URI's in the object position of triples
2. Manually add the missing namespaces for URI's in the object position of triples
3. Assign prefixes to each namespace
4. Use the model.setNsPrefixes(prefixToNamespaceMap) method to update the Model accordingly
5. Write the Model to an RDF string
Is there any way to avoid step 3 by having Jena auto-generate prefixes for namespaces used within the model (including URI's in the object position)? If not, is there an alternative to model.listNameSpaces() that will also return namespaces used in URI's in the object position of triples?
Thanks!
----------------------------------------------------------------------
This message, and any attachments, is for the intended recipient(s) only, may contain information that is privileged, confidential and/or proprietary and subject to important terms and conditions available at http://www.bankofamerica.com/emaildisclaimer. If you are not the intended recipient, please delete this message.
Re: Namespace prefixes in Jena Models
Posted by Andy Seaborne <an...@apache.org>.
On 28/04/16 16:46, Scoggan, Tom wrote:
> Thanks, Andy. Is the answer to this question "no" as well?
>
> Is there an alternative to model.listNameSpaces() that will also return namespaces used in URI's in the object position of triples?
Tom,
"no" though the question is not well defined.
Namespaces exist if and only if they are declared.
In the real data there are no namespaces, all data is URIs.
Namespaces are held in the separate prefix mapping.
Namespaces are applied when writing out the data.
You can list all the objects and calculate a possible namespace URI
using the code in SplitIRI.
Andy
>
> -----Original Message-----
> From: Andy Seaborne [mailto:andy@apache.org]
> Sent: Thursday, April 28, 2016 11:18 AM
> To: users@jena.apache.org
> Subject: Re: Namespace prefixes in Jena Models
> Importance: High
>
> On 28/04/16 15:56, Scoggan, Tom wrote:
>> When writing a Model to an RDF string, I currently do something like this:
>>
>>
>> 1. Use the model.listNameSpaces() method to get the list of namespaces used --- this doesn't include namespaces used in URI's in the object position of triples
>
> Model.listNamespaces list the names set on the model. Either in the
> data when read in, or set by the application.
>
>>
>> 2. Manually add the missing namespaces for URI's in the object position of triples
>
> Namespaces are not necessary - they are a convenience for presentation
> (and that works better in Turtle than RDF/XML). Choosing the prefix or
> namespace name
>
> [except for properties in RDF/XML which can only be written with qnames]
>
>> 3. Assign prefixes to each namespace
>
> Something somewhere is going to have to choose nice prefixes.
>
>>
>> 4. Use the model.setNsPrefixes(prefixToNamespaceMap) method to update the Model accordingly
>>
>> 5. Write the Model to an RDF string
>>
>> Is there any way to avoid step 3 by having Jena auto-generate prefixes for namespaces used within the model (including URI's in the object position)? If not, is there an alternative to model.listNameSpaces() that will also return namespaces used in URI's in the object position of triples?
>
> Set them in the input data.
>
> Or keep a Turtle file that is only prefixes and parse it into any model
> your create. Parsing a file sets the prefixes so if your data is
> N-triples, say, and you want to see it in pretty turtle, just add in
> some extra, domain specific, prefixes.
>
> Andy
>
>>
>> Thanks!
>
> ----------------------------------------------------------------------
> This message, and any attachments, is for the intended recipient(s) only, may contain information that is privileged, confidential and/or proprietary and subject to important terms and conditions available at http://www.bankofamerica.com/emaildisclaimer. If you are not the intended recipient, please delete this message.
>
RE: Namespace prefixes in Jena Models
Posted by "Scoggan, Tom" <to...@baml.com>.
Thanks, Andy. Is the answer to this question "no" as well?
Is there an alternative to model.listNameSpaces() that will also return namespaces used in URI's in the object position of triples?
-----Original Message-----
From: Andy Seaborne [mailto:andy@apache.org]
Sent: Thursday, April 28, 2016 11:18 AM
To: users@jena.apache.org
Subject: Re: Namespace prefixes in Jena Models
Importance: High
On 28/04/16 15:56, Scoggan, Tom wrote:
> When writing a Model to an RDF string, I currently do something like this:
>
>
> 1. Use the model.listNameSpaces() method to get the list of namespaces used --- this doesn't include namespaces used in URI's in the object position of triples
Model.listNamespaces list the names set on the model. Either in the
data when read in, or set by the application.
>
> 2. Manually add the missing namespaces for URI's in the object position of triples
Namespaces are not necessary - they are a convenience for presentation
(and that works better in Turtle than RDF/XML). Choosing the prefix or
namespace name
[except for properties in RDF/XML which can only be written with qnames]
> 3. Assign prefixes to each namespace
Something somewhere is going to have to choose nice prefixes.
>
> 4. Use the model.setNsPrefixes(prefixToNamespaceMap) method to update the Model accordingly
>
> 5. Write the Model to an RDF string
>
> Is there any way to avoid step 3 by having Jena auto-generate prefixes for namespaces used within the model (including URI's in the object position)? If not, is there an alternative to model.listNameSpaces() that will also return namespaces used in URI's in the object position of triples?
Set them in the input data.
Or keep a Turtle file that is only prefixes and parse it into any model
your create. Parsing a file sets the prefixes so if your data is
N-triples, say, and you want to see it in pretty turtle, just add in
some extra, domain specific, prefixes.
Andy
>
> Thanks!
----------------------------------------------------------------------
This message, and any attachments, is for the intended recipient(s) only, may contain information that is privileged, confidential and/or proprietary and subject to important terms and conditions available at http://www.bankofamerica.com/emaildisclaimer. If you are not the intended recipient, please delete this message.
Re: Namespace prefixes in Jena Models
Posted by Andy Seaborne <an...@apache.org>.
On 28/04/16 15:56, Scoggan, Tom wrote:
> When writing a Model to an RDF string, I currently do something like this:
>
>
> 1. Use the model.listNameSpaces() method to get the list of namespaces used --- this doesn't include namespaces used in URI's in the object position of triples
Model.listNamespaces list the names set on the model. Either in the
data when read in, or set by the application.
>
> 2. Manually add the missing namespaces for URI's in the object position of triples
Namespaces are not necessary - they are a convenience for presentation
(and that works better in Turtle than RDF/XML). Choosing the prefix or
namespace name
[except for properties in RDF/XML which can only be written with qnames]
> 3. Assign prefixes to each namespace
Something somewhere is going to have to choose nice prefixes.
>
> 4. Use the model.setNsPrefixes(prefixToNamespaceMap) method to update the Model accordingly
>
> 5. Write the Model to an RDF string
>
> Is there any way to avoid step 3 by having Jena auto-generate prefixes for namespaces used within the model (including URI's in the object position)? If not, is there an alternative to model.listNameSpaces() that will also return namespaces used in URI's in the object position of triples?
Set them in the input data.
Or keep a Turtle file that is only prefixes and parse it into any model
your create. Parsing a file sets the prefixes so if your data is
N-triples, say, and you want to see it in pretty turtle, just add in
some extra, domain specific, prefixes.
Andy
>
> Thanks!