Named entity recognition 2

• looks simple and straightforward

  • use of name lists (morphology!?)

• performance

  • human: 98-99%
  • best systems: 94%

• name identification

  • less problematic

• name categorization

  • more complex
  • needs co-textual clues (e.g. “dr. Pecina”)

Previous slide

Next slide

Back to first slide

View graphic version