1. Start with a small set of seed patterns and seed tuples
    • e.g. seed pattern: /* has a hub at */
    • seed tuple: Ryanair - Charleroi
  2. Search for the terms in the seed tuple, within a corpus of documents
  3. Use the search results to extract new patterns
    • /[ORGANISATION], which uses [LOCATION] as a hub/
    • consider context before, between and after the 2 entities
  4. Search for additional tuples using the new patterns (to grow set of tuples)