Nlao writeup of Krishnan 2006

This is a review of Krishnan_2006_an_effective_two_stage_model_for_exploiting_non_local_dependencies_in_named_entity_recognition by user:Nlao.

This paper nicely analyzes different types of long range dependencies for named entity recognition. It also analyzes in detail how the proposed method correctly capture these dependencies.

Evaluation is solid, uses standard data set, compares with competent methods, and analyzes usefulness of different features.

[minor points] - The proposed method is very much alike to stacked learning, but no connection is drawn.