[HowTo] Recreation of a query

[samwaseda]

I have the following query that I would like to create with tools4RDF:

PREFIX pmd: <https://w3id.org/pmd/co/>
SELECT distinct ?p ?matDesVal
WHERE {
?p pmd:input ?s .
?p pmd:characteristic ?matDes .
?matDes a pmd:materialDesignation .
?matDes pmd:value ?matDesVal .
?s a pmd:TestPiece .
} ORDER BY ?p

Now, I tried the following command

network.create_query(network.terms.co.TestPiece, network.terms.co.value)

and got the result (Note: already the latest PR applied):

PREFIX co: <https://w3id.org/pmd/co/>
SELECT DISTINCT ?TestPiece ?valuevalue
WHERE {
    ?TestPiece co:characteristic ?co_ValueObject .
    ?co_ValueObject co:value ?valuevalue .
    ?TestPiece a co:TestPiece .
}

That looks already pretty nice, but it's still not exactly the same. Quite importantly, I haven't got the column for p yet. How can I get it included?

[srmnitc]

Depends on what p is. Assuming p is co.RandomTerm, you could do network.create_query(network.terms.co.TestPiece, [network.terms.co.RandomTerm, network.terms.co.value])

[samwaseda]

ok in other words without knowing what type p is, we cannot reproduce it?

[srmnitc]

As of now, now. The function assumes you start from a class, and ends at a class or data. I think one could probably tweak the function to start from a predicate, there is no fundamental restriction.

[srmnitc]

What is missing, which might be needed:

• cannot give type as a condition unless the class is explicitely specified
• ORDER by is not implemented

[samwaseda]

Actually I simply copied and pasted @mbruns91's query here, but from my perspective, it's easier for most of the people to deal with the ordering inside pandas, so you are right that it might be good to add it, but I wouldn't be worried even if it's not there.

[samwaseda]

flowchart LR
p(((?p))) -- pmd:input --> ?s
p -- pmd:characteristic --> ?matDes
?matDes -- rdf:type --> pmd:materialDesignation
?matDes -- pmd:value --> m(((?matDesVal)))
?s -- rdf:type --> pmd:TestPiece

Loading

[srmnitc]

As of now, now. The function assumes you start from a class, and ends at a class or data. I think one could probably tweak the function to start from a predicate, there is no fundamental restriction.

@samwaseda have you started on this? if not, I can also give it a try

[samwaseda]

have you started on this? if not, I can also give it a try

Looking at @mbruns91’ examples, I got the feeling that there is a need for a much more generalized approach for this. Do you have a good idea?

Edit: So the answer is no :)

P.S. Sorry for posting the diagram without any comment. I needed it for a discussion with @aazocar

[aazocar]

Hi @samwaseda I'm taking a look at @mbruns91 examples. Quick question: which version of pmdco are you using? https://w3id.org/pmd/co/3.0.0-alpha1 or https://w3id.org/pmd/co/2.0.8 ?

[samwaseda]

I guess @mbruns91 should know it better, but in principle it should be 2.0.8, but if you have tested it you might have noticed that the data and the ontology are not compatible...

[mbruns91]

Hey! I can't check right now, it should be 2.0.8.

[aazocar]

To allow for more flexible queries:

• support optional destination
• source can be class or object property (not data property)
• destination can be class (done), object property and data property.

@srmnitc is working on a draft