"Views" or some similiar mechanism for CSV

[oschihin]

I know, you implemented R2RML Views in the mapping language, that will be useful (https://github.com/zazuko/expressive-rdf-mapper/blob/master/documentation/mapping-language.md#sql-query-as-source). Now, working from home without an easy DB access, I am working with CSV data and RML, respectively.

Here, I wonder if there is some mechanism to choose rows based on a condition. Imagine a source file like:

ID,Name, additionalName,Category
1,Paris, Lutetia,Place
2,Napoleon Bonaparte,Napoleone Buonaparte,Person

and a possible definition for logical sources based on the same file:

source-group Descriptors {
    type csv

    logical-source Places {
        source "descriptors.csv where Category = Place"
    }
    logical-source Persons {
        source "descriptors.csv where Category = Person"
    }
}

In the RML specs, there is no obvious solution for this, maybe the rml:iterator property?

[ktk]

@oschihin there is no such functionality in RML to the best of my knowledge, we have some ideas of doing that in CSV on the Web but right now you need to post-process that somehow on your own. I do that with SPARQL INSERT queries on the data within a CSV on the Web pipeline using bardard59.

Stardog can also load CSV files with R2RML but I'm not sure if you can do SQL on those files.