geopython/pygeometa

Option to describe attributes

pvgenuchten opened this issue · 6 comments

pvgenuchten commented

This is a request to add a section on pygeometa mcf to describe attributes of a dataset.

attributes typically have non-human friendly labels. In metadata you can give a description of attribute names which are localised and human friendly, plus extra details such as type and extra metadata. 4 use cases:

  • option to give localised human friendly label for attributes
  • In the mcf to iso19139 conversion, use the attribute info to convert to a linked iso19110 document (in iso19115:2012, iso19110 can be embedded)
  • for a csv file you have to define a .vrt file for GDAL to indicate which fields to use as Lat and Lon, by including this in the metadata, the vrt can be automatically generated
  • For each attribute a link can be added to a generic ontology, providing a minimal approach for mapping to rdf

Sample:

attributes:
  - name: id
    type: numeric
    identifier: true
  - name: title
    type: string
    href: http://purl.org/dc/terms/title
    title:
      en: title
      fr: titre
  - name: x
    type: geometry/point/lat
  - name: y
    type: geometry/point/lon
tomkralidis commented

Good idea @pvgenuchten. Proposal below:

attributes:
     - name: foo
       title:
           en: foo title
       description:
           en: foo description
       url: https://example.org/foo
       type: int
       units: SI_unit
       values: (enumerated list or min/max)

...where values could be one of:

values: [1, 2, 3, 4]
values:
    range: [1, 4]

Thoughts?

  • 🚀1
pvgenuchten commented

Yes, good suggestion,

Further inspiration can be taken from:

Do we need a thing on cardinality (0,,1..n)
I think we need more guidance on codelist:

  • an option to reference an online codelist like:
values: 
  codelist: 
    title: 
      en: World Reference Soil Base Types
    description:
      en: "The WRB is an international soil classification system under the lead management of FAO for naming soils and creating soil map legends.(en)"
    url: http://aims.fao.org/aos/agrovoc/c_89f35c33
  • or write out the list
values: 
  codelist: 
  - code: "alisol"
    label:
      en: "Alisol"
    description
      en: "Alisols have a higher clay content in the subsoil than in the topsoil, as a result of pedogenetic processes (especially clay migration) leading to an argic subsoil horizon. Alisols have high-activity clays throughout the argic horizon and a low base saturation in the 50–100 cm depth. They occur predominantly in humid tropical, humid subtropical and humid temperate regions. Many Alisols correlate with Parabraunerden (Germany), Argissolos (Brazil), Ultisols with high-activity clays (United States of America), Kurosols (Australia), and Fersialsols and Sols fersiallitiques très lessivés (France)."
    url: http://aims.fao.org/aos/agrovoc/c_479c499a
  - code: "Anthrosol"
    label:
      en: "Anthrosol"
    description
      en: "Anthrosols comprise soils that have been modified profoundly through human activities, such as addition of organic or mineral material, charcoal or household wastes, or irrigation and cultivation. The group includes soils otherwise known as Plaggen soils, Paddy soils, Oasis soils, or Terra Preta de Indio. Many of them correspond to Highly cultivated soils and Anciently irrigated soils (Russia), Terrestrische anthropogene Böden (Germany), Anthroposols (Australia) and Anthrosols (China)."
    url: http://aims.fao.org/aos/agrovoc/c_4515b13e
  • an interesting approach in the table-spec of data package for rdf types (for example skos concepts). Suggestion to define the type of a field as a rdfType, which would automatically indicate how to populate it
tomkralidis commented

@pvgenuchten proposal for values, being one of:

enum

 values:
    enum: [1, 2, 3. 4]

range

 values:
    range: [1, 4]

codelist

values:
    codelist:
     - name: foo
       title:
           en: foo title
       abstract:
           en: foo description
       url: https://example.org/values/foo
pvgenuchten commented

should be #162, giving it a look, thanx Tom

tomkralidis commented

Implemented in #162