part-of-speech (POS) tagger with char+word embedding and BiLSTM

Introduction

This project implements a part-of-speech (POS) tagger. It uses a convolutional neural network for character-level embedding, along with word-level embedding. POS tagging is achieved by bidirectional LSTM. And it is super fast!

Usage

Firstly clone this project from this repo.

To build the tagger

python3.5 buildtagger.py sents.train tagger

To tag corpus

python3.5 runtagger.py sents.test tagger sents.out

To evaluate results

python3.5 eval.py sents.out sents.answer

Contributors

Xudong Shen (xudong.shen@u.nus.edu)