Align Knowledge-Graph node labels to a canonical vocabulary

Computes the mapping from node labels currently present in an edaphos_causal_kg onto their canonical counterparts in a target vocabulary (causal_ontology_cerrado() by default). Three matchers are tried in order — exact, substring, fuzzy — and the first hit wins. The mapping is returned as a tidy data frame; apply it with causal_kg_rename().

Usage

causal_kg_alignment(
  kg,
  vocab = NULL,
  method = c("exact", "substring", "fuzzy"),
  max_distance = 4L,
  agrovoc_cache = NULL,
  agrovoc_batch = FALSE,
  agrovoc_max_active = 5L
)

Arguments

kg

An edaphos_causal_kg.

vocab

Either a character vector / data frame of canonical terms, the string "cerrado" (default; uses causal_ontology_cerrado()) or the string "agrovoc" which triggers a live SPARQL query to the FAO AGROVOC endpoint via causal_ontology_agrovoc_align(). For AGROVOC the alignment type is reported as "agrovoc" instead of exact / substring / fuzzy.

method

Which matcher tier(s) to enable (ignored when vocab = "agrovoc"). Any combination of "exact", "substring", "fuzzy".

max_distance

Fuzzy-matcher Levenshtein cap.

agrovoc_cache

Optional .rds path used by vocab = "agrovoc" to avoid re-querying the same terms.

agrovoc_batch

Logical — when TRUE and vocab = "agrovoc", uses the parallel-dispatch variant causal_ontology_agrovoc_align_batch() to resolve all nodes in flight. Recommended for KGs with more than ~20 unique nodes.

agrovoc_max_active

Integer — concurrency for the parallel variant. Only consulted when agrovoc_batch = TRUE.

Value

A data frame with columns original, canonical, method, distance. When vocab = "agrovoc" an extra uri column is attached.