Ingest an abstract into a KG via multi-extractor voting — causal_llm_ingest_abstract

Single-call equivalent of running causal_llm_vote() and then causal_kg_add_edge() on every surviving edge. The resulting KG has a source tag of the form "<abstract_source> | vote(<voting>, n=<N_backends>)" so the provenance of every edge records both the underlying abstract and the backends that agreed on the claim.

Usage

causal_llm_ingest_abstract_voted(
  kg,
  abstract,
  source,
  backends,
  voting = "majority",
  min_support = NULL,
  threshold = NULL,
  weights = NULL,
  min_confidence = 0.5,
  timeout_sec = 120
)

Arguments

kg, abstract, source: See causal_llm_ingest_abstract().
backends, voting, min_support, threshold, weights, timeout_sec: Forwarded to causal_llm_vote().
min_confidence: Claims whose mean_confidence is below this threshold are dropped before insertion.

Value

The updated edaphos_causal_kg. A claims attribute carries the tidy consensus data frame that was actually inserted, and a per_backend attribute carries the raw per-backend claims for audit / debugging.

Examples

if (FALSE) { # \dontrun{
  backends <- list(
    list(backend = "ollama",    model = "gemma4:latest"),
    list(backend = "openai",    model = "gpt-4o-mini")
  )
  kg <- causal_kg_new()
  kg <- causal_llm_ingest_abstract_voted(
    kg,
    abstract = "Higher MAP drives SOC accumulation in Cerrado...",
    source   = "Ferreira 2021",
    backends = backends,
    voting   = "majority"
  )
} # }