Skip to content

Convolutional neural network and word embeddings for Chinese word segmentation

License

Notifications You must be signed in to change notification settings

chqiwang/convseg

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

33 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CONV-SEG

Convolutional neural network for Chinese word segmentation (CWS). The corresponding paper: Convolutional Neural Network with Word Embeddings for Chinese Word Segmentation

Author

Chunqi Wang

Dependencies

It is better to use a nvidia GPU to accelerate the training procedure.

Data

Downlaod data.zip from here (Note that the SIGHAN datasets should only be used for research purposes). Extract data.zip to this directory. So the file tree would be:

convseg
|	data
|	|	datasets
|	|	|	sighan2005-pku
|	|	|	|	train.txt
|	|	|	|	dev.txt
|	|	|	|	test.txt
|	|	|	sighan2005-msr
|	|	|	|	train.txt
|	|	|	|	dev.txt
|	|	|	|	test.txt
|	|	embeddings
|	|	|	news_tensite.w2v200
|	|	|	news_tensite.pku.words.w2v50
|	|	|	news_tensite.msr.words.w2v50
|	tagger.py
|	train_cws.py
|	train_cws.sh
|	train_cws_wemb.sh
|	score.perl
|	README.md

How to use

First, give execute permission to scripts:

chmod +x train_cws.sh train_cws_wemb.sh

Train a preliminary model (CONV-SEG):

./train_cws.sh WHICH_DATASET WHICH_GPU

Train a model with word embeddings (WE-CONV-SEG):

./train_cws_wemb.sh WHICH_DATASET WHICH_GPU

We have two optional datasets: pku and msr. If you run the program in CPU environment, just leave the second argument empty.

For example, if you want to train the model CONV-SEG on the pku dataset and on gpu0, you should run:

./train_cws.sh pku 0

More arguments can be set in train.py.

Test Score

Model PKU(dev) PKU(test) MSR(dev) MSR(test)
CONV-SEG 96.8 95.7 97.2 97.3
WE-CONV-SEG 97.5 96.5 98.1 98.0

About

Convolutional neural network and word embeddings for Chinese word segmentation

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published