Colpabot

Colpabot is a chatbot developed as part of a bachelor thesis, built using the Microsoft Bot Framework. It is designed to assist patients undergoing colonoscopy preparation.

Features

• User Persistent Data: Initiates conversation by asking for language preference and colonoscopy date, storing this information for personalized interactions.
• Natural Language Processing: Matches user prompts to a predefined Q&A database, applying various operations for accurate responses.
• Image Support: Sends relevant images alongside certain answers.
• Reminders: Provides timely reminders about important aspects, like diet before the colonoscopy.
• Command System: Offers various commands to enhance user interaction. Type /help for more information.
• Multi-platform:
  • Local testing with Bot Framework Emulator
  • Scalable deployment on Azure
  • Integration with Telegram (chat here - Telegram account required)

Prerequisites

• .NET SDK version 6.0

  # Check your dotnet version
  dotnet --version

• Infobip's WhatsApp adapter version 4.13.5 for the integration with WhatsApp

• Fastenshtein version 1.0.10 for Levenshtein distance calculation

• Quartz version 3.13.0 for scheduling reminders

• sqlite-net-sqlcipher version 1.9.172 for encrypting the database

• SixLabors's ImageSharp version 3.1.5 for detecting image sizes

• Optional prerequisites:

  • DeepL's free API key

    After obtaining the key, create a deepl_key.txt file in the same directory as the README and paste the key there.

  • To deploy to Azure, you will need:

    • MicrosoftAppId
    • MicrosoftAppPassword

    These values should be added to appsettings.json. You will also need to modify the BotUrl in this file.

  • For WhatsApp integration via InfoBip, create an InfoBip account and retrieve the following:

    • InfobipApiBaseUrl
    • InfobipApiKey
    • InfobipAppSecret
    • InfobipWhatsAppNumber
    • InfobipWhatsAppScenarioKey

    For more details, refer to this GitHub repository and this documentation.

Running the Bot

• Choose one of the following methods:

  Option A: Command Line

  # Navigate to project folder (ColpaBot)
  cd ColpaBot
  
  # Run the bot
  dotnet run

  Option B: Visual Studio

  • Launch Visual Studio
  • File -> Open -> Project/Solution
  • Navigate to ColpaBot folder
  • Select ColpaBot.csproj file
  • Press F5 to run the project

Testing

Bot Framework Emulator

1. Download and install the Bot Framework Emulator
2. Launch the emulator
3. Go to File -> Open Bot
4. Enter the Bot URL:
   • Usually http://localhost:3978/api/messages when debugging
   • May be http://localhost:5000/api/messages after building and executing
5. If MicrosoftAppId and MicrosoftAppPassword are set in appsettings.json, include them with the Bot URL. These are values obtained from the Azure portal.

Customization

Adding Questions and Answers

1. Navigate to Resources/QuestionsAndAnswers/
2. Open the appropriate language file (e.g., questions_and_answers_EN-GB.tsv for English)
3. Add new rows with questions and answers
   • You can use . to copy the previous answer
   • To attach an image to the message, add ImageName in the actions column
     • This image should be in the Images folder with Name as its file name.
     • To show one of many images with a message, create a folder and with the images you want to show. The names of these images do not matter, but the folder's name should be the Name in the ImageName
   • When using actions, you should take into account the specific rules that the order of actions must follow.
     • Image[...] must be the first one in the general section.
     • CopyQuestion[...] must be the first one in the specific sections. Something to note is that this action copies both the question AND the non-image actions of that entry.
     • DontAnswer must be the last one in all the sections.
4. Save the file
5. Optional: Translate to other languages:
   • Ensure Python and the deepl library are installed
   • Run python translate.py in the script's directory

Modifying Reminders

1. Open Resources/reminders.tsv
2. Follow the file's structure to add or modify reminders
3. For new messages, add them to Resources/bot_messages.tsv
4. Run the translation script

Deployment

The bot was deployed using a GitHub workflow with Visual Studio.

Deploying to Azure

1. Login to Visual Studio with your Microsoft Azure account
2. Go to the project's GitHub repository
3. Click on Actions
4. Select the workflow
5. Click on "Run workflow"

For more details, see Deploy your bot to Azure.

Additional Resources

• Bot Framework Documentation
• Bot Basics
• Azure Bot Service Documentation
• Azure Portal

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

Email