eclipse/rdf4j

https://rdf4j.org/ - discuss general RDF tech, and RDF4J in particular

AF2021

@AF2021

A lot of new terms

AF2021

@AF2021

Is it possible to set RDFXMLPrettyWriter so that it is not nesting statements or is there another thing that has to be changed.

I'm getting this:

<?xml version="1.0" encoding="UTF-8"?>

<rdf:RDF xmlns:dcat="http://www.w3.org/ns/dcat#" xmlns:dcterms="http://purl.org/dc/terms/" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">

<dcat:Catalog rdf:about="http://www.foo">
<rdf:type>
<dcat:Dataset rdf:about="https://example.com/dataset1">
</dcat:Dataset>
</rdf:type>
</dcat:Catalog>

</rdf:RDF>

But I would rather have a flat structure instead like this:

<?xml version="1.0" encoding="UTF-8"?>

<rdf:RDF xmlns:dcat="http://www.w3.org/ns/dcat#" xmlns:dcterms="http://purl.org/dc/terms/" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">

<dcat:Catalog rdf:about="http://www.foo">
<dcat:dataset rdf:resource="https://example.com/dataset1"/>
</dcat:Catalog>

<dcat:Dataset rdf:about="https://example.com/dataset1">
</dcat:Dataset>

</rdf:RDF>

Jeen Broekstra

@jeenbroekstra

@AF2021 there's no direct configuration control over that kind of thing, but the way the RDFXMLPrettyWriter works is that it relies on the order of the RDF statements that put into it. When the subject of a statement is the object of the previous statement (so called 'striped syntax'), it attempts to nest them.

so what you can do to make sure it's not nested is ensure that the statements about dataset1 do not immediately follow the statement where you say that your catalog has a dataset which is dataset1.

fwiw, don't take this as a criticism, but I personally never bother with this kind of thing. RDF/XML syntax is notoriously difficult to work with and to read. If I want something that is easy to read and well-organized I use the Turtle syntax.

Bart Hanssens

@barthanssens

indeed, though there is value in RDF/XML when the rest of the toolchain is XML-based (that's the beauty of RDF, one can express it in different serializations)

Jeen Broekstra

@jeenbroekstra

oh sure, if you have to use RDF/XML for one reason or another, then of course :) I just thought I'd mention that if you have a choice, there are easier syntax formats to work with.

Bart Hanssens

@barthanssens

good point :-) personally I use turtle for taxonomies and N-Triples for all the rest (to do basic queries with unix text tools ;-)... while the webdevs prefer a "framed" JSON-LD ... never used TriG nor TriX but I'm sure there are use cases as well

AF2021

@AF2021

No, that's fine and thanks for the input...I'll test
Maybe turtle is easier.

The file is valid RDF format after all.

Jeen Broekstra

@jeenbroekstra

@AF2021 btw when I say the "order of the RDF statements", it's kind of important to know how you can control that. I think you are using a LinkedHashModel object to create your RDF model, right? If so, you're in luck: a LinkedHashModel keeps the statements in the order in which you put them in. So all you'd need to do to change the order is to switch some of the model.add(....); java operations around.

Jeen Broekstra

@jeenbroekstra

No, that's fine and thanks for the input...I'll test
Maybe turtle is easier.
The file is valid RDF format after all.

:thumbsup: and of course feel free to continue asking questions here

AF2021

@AF2021

@barthanssens Thanks, I appreciate it. And the trick worked :)

@jeenbroekstra Thanks...I appreciate your help :)

Fernando Luján Martínez

@ferchitas

Hello everybody, I'm trying to understand what is the differences among RDF4J and Apache Jena. For me both of them are quite similar, taking into account what is described in the documentation web pages, but it would be great if someone could show me if I'm wrong or not.

Bart Hanssens

@barthanssens

well, they are indeed both java open source libraries / tools voor RDF4j, and both include triple stores, IMHO the main difference is that Jena has better support for reasoning, while RDF4J has excellent support for SHACL validation

as for API, that's probably a matter of taste, personally I prefer RDF4J's API :-)

RDF4J is also supported by various storage tools (like AWS Neptune DB, OntoText GraphDB, Stardog, and quite recently Oracle), though Jena can also be used for e.g. Stardog (and probably a few others, don't know them by heart)

Fernando Luján Martínez

@ferchitas

great, thank you. I'll perform some test with both of them and check which one fits better in our needs. Our main concern is the concurrency accessing to a in-memory dataset (read and write)

Jeen Broekstra

@jeenbroekstra

@ferchitas fwiw RDF4J has full transactional support, which allows for concurrent read/write. See https://rdf4j.org/documentation/programming/repository/#transactions

Bart Hanssens

@barthanssens

@ferchitas you probably want to use the NativeStore if you have > 100 K triples https://rdf4j.org/javadoc/latest/org/eclipse/rdf4j/sail/nativerdf/NativeStore.html

Fernando Luján Martínez

@ferchitas

we are using the repository to convert from a communication interface to other. Inserting the input object and retrieving with a construct sparql a converted object. And we are using Repositories.consume to do this because the documentation is indicated that it is transactional.

@barthanssens we are inserting less than 3k/4k triples in each conversion. When the object has been retrieved the repository is cleared

Bart Hanssens

@barthanssens

ok, good to know, though you still might want to consider to also evaluate the NativeStore, since the MemoryStore does not support concurrent transactions (but perhaps in your case the communication is fast enough to make it a non-issue) https://rdf4j.org/javadoc/latest/index.html?org/eclipse/rdf4j/sail/memory/MemoryStore.html

Fernando Luján Martínez

@ferchitas

The performance is good when there is a few requests, however, when they are increased the component is really slow. I think we could test with the NativeStore.

Jasper Koehorst

@jjkoehorst_gitlab

I was wondering if it is possible to validate an rdf file with rdf4j and encode invalid IRI's on the fly? I have some invalid triples containing ## and trying to figure out if this is possible with rdf4j or that you need an additional encoder?

Bart Hanssens

@barthanssens

validating can be turned on/off via parser settings, but not sure about encoding-on-the-fly part https://rdf4j.org/javadoc/latest/org/eclipse/rdf4j/rio/helpers/BasicParserSettings.html#VERIFY_URI_SYNTAX

there is however a ParsedIRI.create that will do exactly that https://rdf4j.org/javadoc/latest/org/eclipse/rdf4j/common/net/ParsedIRI.html#create-java.lang.String-

AF2021

@AF2021

Hello again,

A new question. I read back an already created .RDF file into a model which looks like this:
<dcat:Catalog rdf:about="http://www.foo"> <dcterms:title xml:lang="sv">Katalogtitel</dcterms:title> <dcterms:description xml:lang="sv">Katalogeskrivning</dcterms:description> <dcterms:publisher rdf:resource="https://example.com/publisher1"/> <dcat:Dataset rdf:resource="https://example.com/dataset1"/> <dcterms:license rdf:resource="http://creativecommons.org/publicdomain/zero/1.0/"/> </dcat:Catalog>

Then I add a statement with the code:

`model.add(cat, DCTERMS.TITLE, valueFactory.createLiteral("apappapapa", "sv"));

and print out the model again and the statement is added like this:

`<?xml version="1.0" encoding="UTF-8"?>

<rdf:RDF xmlns:dcat="http://www.w3.org/ns/dcat#" xmlns:dcterms="http://purl.org/dc/terms/" xmlns:foaf="http://xmlns.com/foaf/0.1/" xmlns:vcard="http://www.w3.org/2006/vcard/ns#" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">

<dcat:Catalog rdf:about="http://www.foo">
<dcterms:title xml:lang="sv">Katalogtitel</dcterms:title>
<dcterms:description xml:lang="sv">Katalogeskrivning</dcterms:description>
<dcterms:publisher rdf:resource="https://example.com/publisher1"/>
<dcat:Dataset rdf:resource="https://example.com/dataset1"/>
<dcterms:license rdf:resource="http://creativecommons.org/publicdomain/zero/1.0/"/>
</dcat:Catalog>

<foaf:Agent rdf:about="https://example.com/publisher1">
<foaf:name xml:lang="sv">Exampel organization</foaf:name>
</foaf:Agent>

<dcat:Dataset rdf:about="https://example.com/dataset1">
<dcterms:title xml:lang="sv">Datasettitel</dcterms:title>
<dcterms:description xml:lang="sv">Dataserbeskrivning</dcterms:description>
<dcat:Distribution rdf:resource="https://example.com/distribution1"/>
<dcterms:publisher rdf:resource="https://example.com/publisher1"/>
<dcat:contactPoint rdf:resource="https://example.com/contactpoint1"/>
</dcat:Dataset>

<dcat:Distribution rdf:about="https://example.com/distribution1">
<dcat:accessURL rdf:resource="http://example.com/api"/>
<dcat:accessService rdf:resource="https://example.com/dataservice1"/>
</dcat:Distribution>

<vcard:Organization rdf:about="https://example.com/contactpoint1">
<vcard:fn xml:lang="sv">Öppna data gruppen på exempel organisationen</vcard:fn>
<vcard:hasEmail rdf:resource="mailto:oppnadata@example.com"/>
</vcard:Organization>

<dcat:DataService rdf:about="https://example.com/distribution1">
<dcterms:title xml:lang="sv">Exempel API</dcterms:title>
<dcat:endpointURL rdf:resource="http://example.com/api"/>
</dcat:DataService>

<rdf:Description rdf:about="http://www.foo">
<dcterms:title xml:lang="sv">apappapapa</dcterms:title>
</rdf:Description>

</rdf:RDF>`

But I would like to have it like this instead.

<dcat:Catalog rdf:about="http://www.foo"> <dcterms:title xml:lang="sv">Katalogtitel</dcterms:title> <dcterms:description xml:lang="sv">Katalogeskrivning</dcterms:description> <dcterms:title xml:lang="sv">apappapapa</dcterms:title> <dcterms:publisher rdf:resource="https://example.com/publisher1"/> <dcat:Dataset rdf:resource="https://example.com/dataset1"/> <dcterms:license rdf:resource="http://creativecommons.org/publicdomain/zero/1.0/"/> </dcat:Catalog>

Is there any way to control this?

Bart Hanssens

@barthanssens

ah, you mean that the added statement is added at the end instead of somewhere at the top ?

AF2021

@AF2021

Yes...sorry about the formatting.

Bart Hanssens

@barthanssens

as Jeen mentioned, you'll probably have to juggle with the order of the statements (or you could write your own XML serializer or apply some XSLT after generating the XML) ...

the prettyprinter can do some reasonable formatting, but may not always be as flexible as one wants in (non-RDF) XML flows

it might work if you first create a Model and (from that model) do some querying/inserting in a new model in the "correct" order

AF2021

@AF2021

Ok, some different options. Since I read my own .RDF file and loop through the statements I need somehow to figure out the correct order. It was easier to put the correct order when the file was first created

Bart Hanssens

@barthanssens

mm yes, well, it's hard to get it "right" and resource-friendly enough for all cases (think circular references, and large files), because RDF/XML basically tries to map a very loose graph structure to a rather strict hierarchical view

Jeen Broekstra

@jeenbroekstra

@AF2021 the BufferedGroupingRDFHandler might be of interest to you. This is a wrapper for Rio writers that takes an input model and makes sure the contents are delivered to the underlying writer in a sensible grouped order. It groups by context, then subject, then predicate. See https://rdf4j.org/javadoc/latest/index.html?org/eclipse/rdf4j/rio/helpers/BufferedGroupingRDFHandler.html

of course a downside of that is that it may break striping (and therefore nesting).

AF2021

@AF2021

@jeenbroekstra The BufferedGroupingRDFHandler did the trick. :) Since I don't want nesting but rather the order I added the statements

AF2021

@AF2021

Assuming I have this this resource which is loaded into the LinkedHashModel:
<dcat:Catalog rdf:about="http://www.foo">
<dcterms:title xml:lang="sv">Katalogtitel</dcterms:title>
<dcterms:description xml:lang="sv">Katalogeskrivning</dcterms:description>
<dcterms:publisher rdf:resource="https://example.com/publisher1"/>
<dcat:Dataset rdf:resource="https://example.com/dataset1"/>
<dcterms:license rdf:resource="http://creativecommons.org/publicdomain/zero/1.0/"/>
</dcat:Catalog>

Is there a simple way to access get the whole "dcat:Catalog" from the model as an aggregate. Is it best to do this with the filter method?

Thanks for all the great help you are providing :)

Bart Hanssens

@barthanssens

hmz, the filter method can be used to get all the statements with a specific subject/predicate and/or object, so you'll get the dcterms:title etc, but not the http://www.foo a <dcat:Catalog> itself

And a minor DCAT note, you probably want to use dcat:dataset (lowercase D) instead of dcat:Dataset (uppercase D) since the property "has dataset" is dcat:dataset...

hmottestad

@hmottestad

Going to release today.

Plan

wait for eclipse/rdf4j#2847
merge develop into master
start release
merge release notes
update maven version numbers

Jeen Broekstra

@jeenbroekstra

Sounds good, go for it!

hmottestad

@hmottestad

release is going well

jenkins is done

now I'm waiting for some PRs to complete CI before I move to updating the maven version numbers and merging the release notes

hmottestad

@hmottestad

@jeenbroekstra Did we agree to move to 4.0.0 now, or stay with 3.x and move to 3.7.0?

hmottestad

@hmottestad

Going to sign off here now. We can decide on 4.0 vs. 3.7 here: eclipse/rdf4j#2853

Jeen Broekstra

@jeenbroekstra

Heads up: we're in the middle of renaming our default branch to main. Some things may not work for a bit and you may need to update your clone. Replacement buses will arrive at platform 6.

Where communities thrive

People

Repo info

Activity