Home

Awesome

Transition-based NER system.

This system is part of a paper accepted at NAACL-HLT 2016 Conference. See the paper here: http://arxiv.org/pdf/1603.01360v1.pdf

Desired labeling

John Smith went to Pittsburgh .
 PER-----   O    O  LOC       O

Corresponding sequence of operations (generated by convert-conll2trans.pl)

SHIFT
SHIFT
REDUCE(PER)
OUT
OUT
SHIFT
REDUCE(LOC)
OUT

Data structures

Operations

Dataset & Preprocessing

We use the datasets from conll2002 and conll2003

Convert conll format to ner action (convert-conll2trans.pl) and convert it to parser friendly format (conll2parser.py).

   perl convert-conll2trans.pl conll2003/train > conll2003/train.trans
   python conll2parser.py -f conll2003/train.trans > conll2003/train.parser 

If you see that the words in the oracle have ' symbol , do the following in the trainind/test/dev datasets:

    []['Peter-NNP', 'Blackburn-NNP'] -->  [][Peter-NNP, Blackburn-NNP]

Link to the word vectors that we used in the NAACL 2016 paper for English: sskip.100.vectors.

Build the system

The first time you clone the repository, you need to sync the dynet/ submodule.

git submodule init
git submodule update

mkdir build
cd build
cmake .. -DEIGEN3_INCLUDE_DIR=/path/to/eigen
make -j2

Training

./lstm-parse -T conll2003/train.parser -d conll2003/dev.parser --hidden_dim 100 --lstm_input_dim 100 -w sskip.100.vectors --pretrained_dim 100 --rel_dim 20 --action_dim 20 --input_dim 100 -t -S -D 0.3 > logNERYesCharNoPosYesEmbeddingsD0.3.txt &

Decoding

./lstm-parse -T conll2003/train.parser -d conll2003/test.parser --hidden_dim 100 --lstm_input_dim 100 -w sskip.100.vectors --pretrained_dim 100 --rel_dim 20 --action_dim 20 --input_dim 100 -m latest_model -S > output.txt
python attach_prediction.py -p output.txt -t conll2003/test -o evaloutput.txt

Evaluation

Attach your prediction to test file

  python attach_prediction.py -p (prediction) -t /path/to/conll2003/test -o (output file)
  ./conlleval < (output file)

Citation

If you make use of this software, please cite the following:

@inproceedings{2016naacl,
  author={Guillaume Lample and Miguel Ballesteros and Kazuya Kawakami and Sandeep Subramanian and Chris Dyer},
  title={Neural Architectures for Named Entity Recognition},
  booktitle={Proc. NAACL-HLT},
  year=2016,
}

License

This software is released under the terms of the Apache License, Version 2.0.

Contact

For questions and usage issues, please contact miguel.ballesteros@upf.edu