• Installation
  • Prerequisites
  • Virtual environment
  • Dependencies
  • Environment variables
• Usage
  • Download data
  • EDA (Exploratory Data Analysis)
  • Create a LUIS app in the LUIS portal
  • Train LUIS Model
  • Test and debug with the Emulator
  • Tutorial: Provision a bot in Azure
  • Deploy Bot to Azure Web App
• Quality Assurance
• Troubleshooting

Repository of OpenClassrooms' AI Engineer path, project #10 .

Goal : use Azure Cognitive Services (LUIS), Azure Web App and Azure Application Insights, to build a flights booking chatbot, integrate it in a web application, and monitor its quality.

You can see the results here :

• Presentation
• Main notebook
• Datataset profile report

This is the project architecture in production :

• Integrate model output into a finished product :
  • Azure Language Understanding - LUIS : train and setup a dedicated language understanding model
  • Azure App Service : deploy an interactive API
  • Azure Bot - Test in Web Chat : test the bot in a web chat
• Integrate an AI processing chain into an IT tool using a code version management tool :
  • GitHub repository : manage the code
  • Automated tests : test the bot
• Control the performance of the model in production
  • Azure Application Insights : monitor the quality of the bot

• Python 3.9
• Azure Cognitive Services - LUIS

# python -m venv env
# > or just :
make venv
source env/bin/activate

# pip install -r notebooks/requirements.txt -r bot/requirements.txt
# > or :
# pip install -r requirements.txt
# > or just :
make install

Set environment variable values in (copy or rename .env.example) :

• .env
• bot/.env

Download, and extract the Language Understanding Model (LUIS) training files from Frames Dataset :

make dataset

The main notebook presents the result of the EDA (exploratory data analysis) :

jupyter-lab notebooks/main.ipynb

Follow the official documentation : Create a LUIS app in the LUIS portal

The luis notebook formats the data, runs the LUIS training and tests the model :

jupyter-lab notebooks/luis.ipynb

Run the bot locally :

make bot-start

Follow the official documentation : Test and debug with the Emulator

Follow the official documentation : Tutorial: Provision a bot in Azure

Build and deploy the bot to Azure Web App integrating the Bot Service, LUIS and Application Insights :

make bot-deploy

# make isort
# make format
# make lint
# make bandit
# make mypy
# make test
# > or just :
make qa

This will run tests on the bot tests/test_bot.py :

• test the LUIS service integration
• test a dialog where the bot gathers the flight informations from the user
• test a dialog where the user gives all the flight informations at once

No known issues...