Skip to contents

Decide a semantic review slot

Source: R/review-console.R
accept_suggestion.Rd

accept_suggestion() records that a candidate is the right term for a slot; reject_suggestion() records that none is, and clears the field. Both are pipe-friendly – they take a review and return it – so a whole review is an ordinary, re-runnable R script. Nothing is written until apply_sdp_semantics() is called.

Usage

accept_suggestion(
  review,
  column = NULL,
  role,
  rank = 1L,
  table = NULL,
  code_value = NULL,
  iri = NULL
)

reject_suggestion(
  review,
  column = NULL,
  role,
  table = NULL,
  code_value = NULL,
  reason = NULL
)

Arguments

review

An ms_semantic_review object from review_semantics().

column

Column name of the slot. Omit it for a table-level slot (tables.csv - observation_unit_iri), which has no column; pass table instead. review_semantics() prints the right spelling either way.

role

Semantic role of the slot ("variable", "property", "entity", "unit", "constraint", "statistical_modifier").

rank

Rank of the candidate to accept, as printed in the shortlist.

table

Table identifier; needed only when the column name appears in more than one table.

code_value

Code value; needed only for code-level slots. Pass "" (or NA) to select a column's own slot when codes of that column have slots with the same role, as a measurement column's codes do: leaving code_value out matches those code slots too. A blank never selects a code's slot, even for a codes.csv row that leaves code_value empty because it supplies vocabulary_iri. review_semantics() prints it whenever it is needed.

iri

Optional IRI to accept instead of a shortlisted candidate – for the case where the right term exists but retrieval did not surface it. An iri that a shortlisted candidate in the slot carries is recorded as that candidate, exactly as its rank would be.

reason

Optional free-text reason recorded with a rejection.

Value

The review, with the decision recorded.

Details

These are the calls review_semantics() prints. Pasting the printed line is the intended workflow, and it is what makes the decision reproducible: the script is the audit trail.

See also

review_semantics(), apply_sdp_semantics()

Examples

dict <- tibble::tibble(
  dataset_id = "demo-1", table_id = "spawners",
  column_name = "spawner_count", term_iri = NA_character_
)
attr(dict, "semantic_suggestions") <- tibble::tibble(
  dataset_id = "demo-1", table_id = "spawners",
  column_name = "spawner_count", code_value = NA_character_,
  dictionary_role = "variable", target_scope = "column",
  target_sdp_file = "column_dictionary.csv", target_sdp_field = "term_iri",
  target_row_key = "demo-1/spawners/spawner_count",
  label = "Spawner Abundance", iri = "https://w3id.org/smn/SpawnerAbundance",
  source = "smn", ontology = "smn",
  definition = "Mature salmon returning to spawn.", score = 4.9
)

review <- review_semantics(dict)
review <- accept_suggestion(review, "spawner_count", "variable", rank = 1)
review$decision
#> [1] "accept"