dongdonglotty / Intent-Classification-Baseline-System

A baseline system for intent recognition using the dataset released by SMP2018 ECDT Shared Task

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Intent Classification Baseline

中文语音助手意图识别基线系统,外部专有名词知识、训练模型和预测样本的流程。

By W. Y. Shen

PROJECT:https://github.com/GhibliField/Intent-Classification-Baseline-System

USE

DOWNLOAD PROJECT

git clone https://github.com/GhibliField/Intent-Classification-Baseline-System.git

DEPENDENCIES

pip install -r requirements.txt

EXAMPLE

RUN SERVER:

python ServerDemo.py 

TEST:

python test.py 

PROJECT STRUCTURE

FILES

├── config.py
├── data
│   ├── devel.json
│   └── train.json
├── models
│   ├── ltp_data_v3.4.0/
│   ├── some.pkl
│   └── ...
├── README.md
├── RequestHandler.py
├── requirements.txt
├── ServerDemo.py
├── test.py
├── trainers
│   ├── classifiers.py
│   ├── NB-SVM_with_chi2.ipynb
│   ├── SVM_epg-tvchannel.ipynb
│   ├── SVM_video-cinemas.ipynb
│   ├── SVM_video-epg.ipynb
│   ├── SVM_video-music.ipynb
│   └── SVM_website-app.ipynb
└── utils
    ├── predicates.tsv
    ├── pretty_sure.tsv
    ├── proper_noun/
    ├── proper_noun_all.tsv
    ├── rules.tsv
    ├── stopword_2792.txt
    └── utils.py

MAIN COMPONENTS

  • data: 训练数据
  • model: 持久化了的分类模型与必要的对象
  • trainer: notebook脚本,用于训练不同的分类器
  • utils: 外部词库,方法等
  • config.py: 配置路径参数, 请务必修改路径
  • RequestHandler.py: 测试流程
  • ServerDemo.py: Flask 测试服务端
  • test.py: 测试脚本

REFERENCE

About

A baseline system for intent recognition using the dataset released by SMP2018 ECDT Shared Task


Languages

Language:Jupyter Notebook 66.6%Language:Python 33.4%