extended-embedding-pub

Sentiment Analysis by radical(of Japanese kanji character)-level encoder.

This encoder refers to following paper.

• https://www.jstage.jst.go.jp/article/tjsai/33/4/33_D-I23/_pdf/-char/ja)

Environment

• OS: Ubuntu 16.04.4 LTS
• GPU: GeForce GTX 1080
• Python: 3.6.5
• Chainer: 4.2.0
• MeCab: 0.996 （split sentence to words）

Usage

Directory configuration as follows.

├── create_vocab.py
├── radical_classifier.py
├── char2radical.py
└── datasets
    ├── radical_dic.csv
    └── set [train/test datasets]

Make vocabulary-dictionary of your datasets with the ./datasets/radical_dic.csv

python create_vocab.py

Edit Hyper-parameter, Run training script with the generated file.

python radical_classifier.py

Licence

MIT

Author

syunna5