Extractors

Extractors are responsible for identifying and creating facts from source content.

Available Extractors

• ManualExtractor - API-driven fact creation
• LLMExtractor - AI-powered extraction
• RuleBasedExtractor - Pattern matching

Base Class

All extractors inherit from FactDb::Extractors::Base:

class Base
  attr_reader :config

  def initialize(config = FactDb.config)
    @config = config
  end

  def extract(content)
    raise NotImplementedError
  end

  def extraction_method
    self.class.name.split("::").last.sub("Extractor", "").underscore
  end
end

Creating Custom Extractors

class MyExtractor < FactDb::Extractors::Base
  def extract(source)
    facts = []

    # Your extraction logic
    # Parse source.content
    # Create fact records

    facts
  end
end

Using Extractors

Via FactDb

facts = FactDb.new
extracted = facts.extract_facts(source.id, extractor: :llm)

Directly

extractor = FactDb::Extractors::LLMExtractor.new(config)
facts = extractor.extract(source)

Extractor Selection

Extractor	Best For	Accuracy	Speed
Manual	High-stakes facts	Highest	Slowest
LLM	Complex documents	High	Medium
Rule-based	Structured content	Medium	Fastest