Re: Parameters for metrics - Issue-223

Hi Antoine,

I must have skipped the previous emails regarding this issue - apologies.


> In the end, what we want for metric parameters is properties like :onLanguage used with specific metrics as in the following:
> [
> :linksetImportingENSKOSPrefLabel  rdf:type dqv:Metric ;
>          skos:definition " Importing on English PrefLabel"@en;
>           dqv:inDimension :completenessOfComplemented ;
>           :onProperty :skosPrefLabel ;
>           :onLanguage <http://id.loc.gov/vocabulary/iso639-1/en> .
> ]
> (see ongoing efforts and examples at [2])
> 

I am not sure if I am understanding your problem correctly, but for me this sounds like you would like to create some kind of restriction on the metric assessment (i.e. checking that the language of the value attached to a skos:prefLabel property has the english language). Am I right? If so, this is not related to daq:requires property. 

Just to give you an example of how daq:requires can be used in a typical case:

—— T-BOX —— 
:DataCubeCompleteness  rdfs:subClassOf daq:Metric;
   daq:expectedDataType xsd:double .

—— A-BOX ——
:CompanyXDCCompleteness  a :DataCubeCompleteness ;
  daq:requires <http://example.org/codelist> ;
  daq:hasObservation ……

So in theory, the requires property is usually used in the A-BOX (as different metrics would have different ‘values’ for the requires property), rather that the T-BOX as in your example. I agree that the daq description is vague and must be improved.



> Our problem is that is just no way of linking :onLanguage with daq:requires that seems satisfactory. There are indeed two 'alignment' options:
> 
> 1. If we define :onLanguage as a sub-property of daq:requires, then we get statements like
> [
> :linksetImportingENSKOSPrefLabel :requires <http://id.loc.gov/vocabulary/iso639-1/en> .
> ]
> which is very confusing for us.
> 
> 2. If we try to create a (sub) classes of Metric that 'requires' the property :onLanguage like
> [
> :MetricWithLanguageParameter rdfs:subClassOf  dqv:Metric ;
>    daq:requires :onLanguage .
> ]
> This is more intituive object for a daq:requires statement. But it is not compatible with the current domain of daq:requires, which is set to be daq:Metric and not rdfs:Class.
> 
> So unless one can solve the issue for us, we will not seek alignment between whatever solution we have for Issue-223 and daq:requires.

I think both of these alignment options are ok (actually the 2nd one looks good and compatible), but I think that you are looking for something different than daq:requires in this case.

This kind of restrictions are really interesting, but we didn’t really consider them in daQ directly - we are actually defining our own quality metric language to define metrics.

I really hope this helps a bit

Cheers,
Jer

Received on Wednesday, 16 March 2016 14:05:16 UTC