t-will19 / bi-lstm-char-ner

A fork edit version of https://github.com/Determined22/zh-NER-TF By add Other Features and Char embedding layer on English Corpus

Bi-LSTM-CHAR-NER

A fork edit version of https://github.com/Determined22/zh-NER-TF By add Other Features and Char embedding layer on English Corpus


View Demo

Table of Contents

• About the Project
  • Built With
• Getting Started
  • Installation
• Usage
• Roadmap
• Contributing
• License
• Contact
• Acknowledgements

About The Project

NER is a basic task in NLP. This a fork edit version of https://github.com/Determined22/zh-NER-TF By add Other Features and Char embedding layer

This project use dataset download from https://github.com/kamalkraj/BERT-NER data dir. Because the edit is simple, only release two jupyter notebooks for record.

By add Char embedding layer, The model accuracy improve from average 70% to 80% training from scratch.

Built With

• TensorFlow 1

Getting Started

Installation

Install TensorFlow 1 and requirements according to https://github.com/Determined22/zh-NER-TF

Usage

1. Download Data from https://github.com/kamalkraj/BERT-NER/tree/dev/data

2. Generate features like follows

-DOCSTART- -X- -X- O

 

EU NNP B-NP B-ORG

rejects VBZ B-VP O

German JJ B-NP B-MISC

call NN I-NP O

to TO B-VP O

boycott VB I-VP O

British JJ B-NP B-MISC

lamb NN I-NP O

. . O O

 

Peter NNP B-NP B-PER

Blackburn NNP I-NP I-PER

 

BRUSSELS NNP B-NP B-LOC

1996-08-22 CD I-NP O

 

1. Data Process and Network Train

2. Model Validation and conclusion decode

Roadmap

Contributing

License

Distributed under the MIT License. See LICENSE for more information.

Contact

svjack - [email protected]

Project Link: https://github.com/svjack/Bi-LSTM-CHAR-NER

Acknowledgements