Customize a robot to help you chat with him

Auto chat example

Chat 1:

user: 在吗？
bot: 在
user: 在干嘛呢？
bot: 看电视
user: 看啥电视呀
bot: 活色生香
user: 很好看吗？
bot: 特搞笑
user: 你在哪里呀？
bot: 家里
user: 家里就你一个人嘛？
bot: 我喜欢一个人玩
user: 那我过来找你？
bot: 不可以，乖乖上班去

Chat 2:

Q: 在吗？
A: 在
Q: 干嘛呢？
A: 没事
Q: 陪我去逛街嘛？
A: 嗯
Q: 你在打游戏？
A: 没有
Q: 那去不去？
A: 去

This is based on the training of 2 million chat records, you can use the records of yourself and your girlfriend to try the effect: P

As for how to use the GitHub robot, you can search for it at 0615fc7eca5cb4

project instruction

chatbot is a Go Q&A engine that quickly generates answers from known dialog data sets.

Why is there a chatbot project?

Many years ago, when we needed a chat bot, I first used ChatterBot , but after using it, it takes about 21 seconds to answer a question with our 120 million paired speech data training model, which is really unacceptable. After carefully reading the ChatterBot , I re-implemented one with Go and optimized it in parallel with go-zero's MapReduce framework. As a result, our answer took about 18 milliseconds on average.

During the National Day holiday, I had a little free time, so I organized this project to open source. One is to give you an actual go-zero MapReduce example; the other is to provide you with a chat robot project to play with.

BTW: I may open source the smart customer service robot project in the future, you can follow my github:

https://github.com/kevwan

Code directory and command line instructions

bot

Q&A engine, you can customize your own matching algorithm

cli

• train

  Train the given question and answer data and generate .gob file

  • -d reads all json and yaml corpus files in the specified directory
  • -i reads the specified json or yaml corpus file, multiple files are separated by commas
  • -o specifies the output .gob file
  • -m timing print memory usage

• ask

  A sample Q&A command line tool

  • -v verbose
  • -c trained .gob file
  • Several possible answers for -t

Data Format

If you have corpus data, you can organize it yourself for training.

The data format can be provided through yaml or json file, https://github.com/kevwan/chatterbot-corpus refer to the format in 0615fc7eca5fb8. It is roughly as follows:

categories:
- AI
conversations:
- - 什么是ai
  - 人工智能是工程和科学的分支,致力于构建具有思维的机器。
- - 你是什么语言编写的
  - Python
- - 你听起来像机器
  - 是的,我受到造物者的启发
- - 你是一个人工智能
  - 那是我的名字。

Thanks

go-zero - https://github.com/zeromicro/go-zero

go-zero in the core/mr package of MapReduce chatbot the response efficiency of 0615fc7eca6012!

ChatterBot - https://github.com/gunthercox/ChatterBot

At first I used ChatterBot , but because the answer was too slow, I could only implement it myself. Thanks to ChatterBot , a great project!

project address

https://github.com/kevwan/chatbot

Welcome to use and support star