Nlp>> vntk-tagger>> 返回
项目作者： vntk

项目描述 ：
Experiments of Basic Vietnamese NLP Problems and Named Entity Recognition Tool
高级语言： JavaScript
项目地址: git://github.com/vntk/vntk-tagger.git
创建时间： 2017-11-21T05:09:02Z
项目社区：https://github.com/vntk/vntk-tagger
开源协议：MIT License
下载


@vntk/tagger

Vietnamese Named Entity Recognition Tool

Join the chat at https://gitter.im/vntk/vntk-tagger

Install using npm:

npm install -g @vntk/tagger

Two main features:

  • Train a new NER from raw data, like: News, QA, Comments, or Chat logs.
  • Predict new text input into entities: PER, ORG, LOC, DATE, TIME, …

Predict new text input

Simply run following command to predict new input from file:

vntk-tagger predict [your_file_name.txt]

The output is a new file with name: your_file_name.txt.tags

Train a new NER from raw data

Step 00 - Crawl new data

Preparing your data from: News, QA, Comments, or Chat logs.

Step 01 - Cleaning

Convert raw data to enrich data, by:

  • remove junk characters
  • remove comments
  • delete empty line

Command: vntk-tagger clean [your_data_file.txt]

Step 02 - Preprocessing

Run node preprocess.js to clean and convert raw data to iob format. Result can feed to the trainer!

  • Convert xml to iob format
  • Tag raw input to

Step 03 - Training

Run the following command to train new NER data.

vntk-tagger train [your_training_file.txt]