Skip to contents

Extract resources from text with an LLM

Source: R/extract_llm.R
extract_llm.Rd

Sends the text to the configured provider and parses the returned JSON into candidate resources. Requires an API key (from the environment); returns an empty list if none is configured or the call fails.

Usage

extract_llm(text, llm = krt_llm(), ...)

Arguments

text

Manuscript text.

llm

A krt_llm() configuration.

...

Reserved.

Value

A list of krt_resource candidates.

Examples

# Uses a mock provider so no network or key is needed:
register_llm_provider("mock",
  function(prompt, llm) '[{"resource_type":"Software/code","display_name":"R"}]')
extract_llm("text", krt_llm("openai"))  # empty without a key
#> list()
extract_llm("text", structure(list(provider = "mock"), class = "krt_llm"))
#> [[1]]
#> <krt_resource> [Software/code] R
#>