Wka writeup of Klein and Manning 2002

From Cohen Courses
Jump to navigationJump to search

This is a review of klein_2002_conditional_structure_versus_conditional_estimation_in_nlp_models by user:wka.

Some Notes:

  • Problem domains:
    • WSD studied with NB model, with JL and CL estmation: discriminative criteria lead to better test-set accuracy
    • POS tagging studied with HMMs and CMMs
  • Label bias and observation bias:
    • explaining-away phenomena
    • consequences of independence assumptions

Comments:

  • (+1) new results consistent with previous research, with generalized conclusions