Skip to content

Latest commit

 

History

History
92 lines (74 loc) · 5.73 KB

README.en.md

File metadata and controls

92 lines (74 loc) · 5.73 KB
Raw

ChatGPT-Customer-Service

中文 | English

license Release

Introduction

This project demonstrates how to build an intelligent customer service chatbot using the FAQ from Taoyuan Airport. The data can be replaced and modified as needed.

Demo

Installation

Obtaining Tokens

  1. Obtain the API Token from OpenAI:
    1. Register/login to an account on the OpenAI platform
    2. Click on the avatar in the upper right corner and select View API keys
    3. Click onCreate new secret key in the middle -> the generated key is the OPENAI_API_KEY (which will be used later)
    • Note: Each API has a free quota and restrictions, see OpenAI Pricing for details.
  2. Obtain the Line Token:
    1. Login to Line Developer
    2. Create a bot:
      1. Create a Provider -> Click Create
      2. Create a Channel -> Select Create a Messaging API channel
      3. Enter the required basic information
      4. After completing the input, there is a Channel Secret under Basic Settings -> Click Issue, the generated key is the LINE_CHANNEL_SECRET (which will be used later)
      5. Under Messaging API, there is a Channel access token -> Click Issue, the generated key is the LINE_CHANNEL_ACCESS_TOKEN (which will be used later)

Project Setup

  1. Fork the Github project:
    1. Register/login to a GitHub account
    2. Go to ChatGPT-Customer-Service
    3. Click Star to support the developer
    4. Click Fork to copy all the code to your repository
  2. Deploy (free space):
    1. Go to replit
    2. Register/login to an account
    3. Fork the ChatGPT-Customer-Service-HyperDB

Project Execution

  1. Environment variable settings
    1. After completing the Fork in the previous step, click on Secrets on the left bottom corner of the project management page in Replit.
    2. Click Got it on the right to add environment variables, including:
      1. The selected model:
        • key: OPENAI_API_KEY
        • value: [Obtained from step 1]
      2. Line Channel Secret:
        • key: LINE_CHANNEL_SECRET
        • value: [Obtained from step 1]
      3. Line Channel Access Token:
        • key: LINE_CHANNEL_ACCESS_TOKEN
        • value: [Obtained from step 1]
  2. Execution:
    1. Click on Run on the top
    2. After successful execution, the message Hello World will be displayed on the right side of the screen, and the URL on the top of the screen should be copied
    3. Go back to the Line Developer, paste the URL obtained in the previous step above Webhook URL under Messaging API, and add /callback. For example: https://ChatGPT-Customer-Service-HyperDB.explainthis.repl.co/callback
    4. Open Use webhook below
    5. Turn off Auto-reply messages below
    • Note: If there is no request within one hour, the program will stop, so the next step is necessary.
  3. CronJob sends requests regularly
    1. Register/Login to cron-job.org
    2. Select CREATE CRONJOB in the upper right corner of the dashboard
    3. Enter ChatGPT-Line-Bot in the Title and enter the URL from the previous step. For example: https://ChatGPT-Customer-Service-HyperDB.explainthis.repl.co/
    4. Set to run every 5 minutes below
    5. Click CREATE

How to customize and modify?

Currently, the project will read the taoyuan-airport-faq file, which has the following format:

category question answer
違禁品 安檢時碰到不得上機之危險（安）物品及違禁品，應如何處理？ 不得上機之危險（安）物品﹕例如超過100ml容器液體、防風型打火機、水果刀…等），需丟棄或改託運；違法違禁品（槍、武士刀、扁鑽、毒品…等），將遭航空警察局移送法辦。
違禁品 請問樂器該如何上飛機? 大型樂器需託運，小型可置於機上隨身物品放置箱者可隨身攜帶。
違禁品 請問出境安檢手提行李的規定 （刮鬍刀、刮鬍泡... 出境安檢手提行李規定：1. 電動刮鬍刀可以攜帶登機，丟棄式刮鬍刀、摺疊式刮鬍刀等有刀片外露形式者，禁止攜帶登機。 2. 刮鬍泡、容器...

You can directly modify this file with the same format. After modifying the file, remember to delete the hyperdb.pickle.gz file that was originally on the Replit website before running it again. The purpose of this file is to pre-process and store the file in Embedding. Therefore, when switching to a new dataset, the old file needs to be deleted.

What if the data is not in CSV format?

You can try modifying the data_preprocessing function in src/utils.

Support Us

If you like this project and would like to support us, you can buy us a coffee. This will be the driving force for us to continue moving forward!

Buy Me A Coffee

Related Projects

License

MIT