Wiki: Lecture 6 - Dependency Parsing


#1

Lecture video

Lecture notes

Discussion Points

  • Syntactic Structure: Consistency and Dependency

    • What information does syntactic structure provide and how can such information help in NLP tasks?
    • What are the components of a syntactic parser and when is it used in a typical NLP pipeline?
    • Why do we still build (and use) syntactic parsers when we can instead try to train an end-to-end RNN and let it figure out what syntactic structure is useful for the task?
    • See slides on inductive bias
  • Dependency Grammar

    • What were the motivations behind context free grammars (CFGs) and what issues do they pose for real-life NLP tasks?
    • How are dependency grammars different from CFGs and how are they better for real-life NLP tasks?
  • Transition-based dependency parsing

    • What was Nivre’s contribution to dependency parsing algorithms?
  • Neural dependency parsing

Assignment:

  • Start working on Assignment 2 (Requires knowledge in tensorflow).
  • Create a thread containing your answers of the discussion points above.