Wende(问得) Chinese Question Answering, is a small and simple factoid chinese question answering system written in Python. You ask a chinese natural language question and Wende will try to give a certain answer of the question. This is still a work in progress, currently the system can only judge the type of the question that the user ask, which means, only the question classification module has been implemented.
-
Install and startup Docker Desktop.
-
Clone the source.
$ git clone https://github.com/h404bi/wende $ cd wende/ -
Download the models and place them into their coresponding directories.
- LTP model: http://ltp.ai/
- Wende & Word2Vec: https://pan.baidu.com/s/1nv5ubJr (1dbm)
P.S. you can also train Wende model by yourself, see
wende/classification/model.pyfor more information. -
Then execute
docker-composeinside the directory, to build Docker image.$ docker-compose up
-
Waiting for image building, it will take some times, after building success, open your browser and enter
127.0.0.1:9191, boom! -
Play with it.
Code: MIT License, Copyright (c) 2016-present, by Chawye Hsu
Data: CC BY-NC-SA 4.0, except for the ltp model use its original license.
This is just the graduation project of my bachelor's degree. (这只是我的学士学位毕业设计项目)