quick history of nlp

Tags: nlp

• 1960’s - pattern matching and rules, unrealistic expectations

  • very large systems attempting symbolic representations

• 1970’s-80’s - linguistically rich, logic-driven systems, labor-intensive successes on few narrow tasks

  • working dialogue systems
  • heavily engineered for specific domains

• 90’s-20’s - statistical revolution, machine learning becomes a central part of nlp

• 2010’s - deep learning takes off and accelerates progress

  • 2018-2019 - new surge of progress on unlabled data

• result, progress is very fast, months to one year on lead times

  • -> focus on metalearning, how to learn quickly