Provide specification for CSV/TSV

[cmungall]

Continued from #30 (comment)

From:
https://docs.google.com/document/d/1xnmLNRMLO7796VmI4GEVv1jadY3e3srLD3gF03h2GqU/edit#heading=h.97ugvf88bemn

Headers:
Defined class: 2 columns: name; or curie.
Named by var: 2 columns per var - name; or curie
String list vars: delimiter should be '|' NOT ','
Override columns:
For name: New column or re-use defined class label column.
Decision: add column: "override label". If filled out, over-rides auto-generated name.
Def: New column
TODO: Add check to prevent re-use of def column name in var.

See also: https://www.w3.org/TR/tabular-metadata/

[dosumis]

@cmungall @balhoff : A first attempt at a spec using https://www.w3.org/TR/tabular-metadata/ :

https://github.com/dosumis/dead_simple_owl_design_patterns/blob/master/spec/dosdp_tsv_schema.md

Comments please.

One question not addressed: How do we associate spreadsheets with specific patterns. Would a naming convention be too lightweight? Or should we have additional mapping files / some way of embedding this as metadata? With that we could specify a curies too.

[dosumis]

Couple of other questions:

One other potential issue, should the type of the ID field be

https://www.w3.org/TR/xmlschema11-2/#anyURI

(i.e. does this cover curies?)

[dosumis]

@cmungall @balhoff - ping.

[balhoff]

Re: the type of the ID field, if you want CURIEs then I think it's more like what they call CURIEorIRI here: https://www.w3.org/TR/rdfa-core/#dfn-curieoriri