GROUP_CONCAT with OPTIONAL returns 'None'

[KonradHoeffner]

A query such as the following on https://hitontology.eu/sparql returns empty strings for nonexisting GROUP_CONCAT results using Virtuoso:

PREFIX dce: <http://purl.org/dc/elements/1.1/>
PREFIX hito: <http://hitontology.eu/ontology/>
SELECT
?classified
(GROUP_CONCAT(DISTINCT(STR(?dce_source));separator="|") AS ?dce_sources)
{
 ?classified a hito:FeatureClassified.
 OPTIONAL {?classified dce:source ?dce_source.}
}

However when querying an RDFLib graph, I get rdflib.term.Literal('None') values in my result set. Is that intended?

[aucampia]

@KonradHoeffner could you give some more detail on this? What was the actual graph you were querying or were you using a remote store?

[KonradHoeffner]

I am using the default graph with a remote SPARQL endpoint. Here is a complete minimal example of the problem:

RDFLib Example Code

import rdflib
g = rdflib.Graph()
qres = g.query(
 """   
PREFIX dce: <http://purl.org/dc/elements/1.1/>
PREFIX hito: <http://hitontology.eu/ontology/>
                                                                                               
SELECT ?classified (GROUP_CONCAT(DISTINCT(STR(?dce_source));separator="|") AS ?dce_sources)
{                                                  
    SERVICE <https://hitontology.eu/sparql>
    {
        ?classified a hito:FeatureClassified.
        OPTIONAL {?classified dce:source ?dce_source.}
    }
}    
LIMIT 3    
"""    
)
for row in qres:
    print(row)  

RDFLib Output

$ python rdflibtest.py
(rdflib.term.URIRef('http://hitontology.eu/ontology/BbAdministrateTheMasterPatientIndexMpi'), rdflib.term.Literal('None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None|None'))
$ pip install rdflib
Defaulting to user installation because normal site-packages is not writeable
Requirement already satisfied: rdflib in /home/konrad/.local/lib/python3.10/site-packages (6.1.1)
Requirement already satisfied: setuptools in /usr/lib/python3.10/site-packages (from rdflib) (59.5.0)
Requirement already satisfied: isodate in /home/konrad/.local/lib/python3.10/site-packages (from rdflib) (0.6.1)
Requirement already satisfied: pyparsing in /usr/lib/python3.10/site-packages (from rdflib) (3.0.8)
Requirement already satisfied: six in /usr/lib/python3.10/site-packages (from isodate->rdflib) (1.16.0)

Manual Query on https://hitontology.eu/sparql

PREFIX dce: <http://purl.org/dc/elements/1.1/>
PREFIX hito: <http://hitontology.eu/ontology/>
                                                                                               
SELECT ?classified (GROUP_CONCAT(DISTINCT(STR(?dce_source));separator="|") AS ?dce_sources)
{                                                  
        ?classified a hito:FeatureClassified.
        OPTIONAL {?classified dce:source ?dce_source.}
}    
LIMIT 3  

Manual Query Result

classified	dce_sources
http://hitontology.eu/ontology/EhrSfmPatientspecificMedicationDosingAndWarnings
http://hitontology.eu/ontology/EhrSfmUserAccessSuccessfulSecurityAuditTrigger
http://hitontology.eu/ontology/EhrSfmUserPermissionsAuthorizationSecurityAuditTrigger

Comparison

RDFLib generates "None|None|None|..." but Virtuoso SPARQL generates an empty string as intended.

[aucampia]

Thanks a bunch for the detailed info, very helpful, will see what I can do about it.

[ghost]

fwiw, adding a GROUP BY ?classified (as appearing in the Virtuoso optimized query), changes the number of None results to just three.

[aucampia]

I can more or less reproduce this, but should GROUP_CONCAT not be accompanied with a GROUP BY clause? It seems Virtuoso is fine with no GROUP BY but Jena does not like it.