MAIR-G14

Welcome to the repository for our MAIR project. This project is a chatbot which can help you find a restaurant.

Functionality

The chatbot can help you find a restaurant based on the following criteria:

• Location
• Cuisine
• Price range

The chatbot will follow the following state-transition diagram to retrieve your preferences and provide you with a restaurant recommendation:

Installing and running

The following instructions should work on any system running Python version 3.10 and higher.

1. Open the root directory of the project MAIR-G14.
2. To install the required packages, run pip install -r requirements.txt.
3. Run python src/main.py to start a conversation.

If you want to use the speech recognition feature, you need to follow the following additional steps if your OS is listed below:

Mac

1. brew install portaudio
2. brew install flac
3. pip install pyaudio

Configuration

The chatbot can be configured by adding arguments to your python src/main.py command. The following arguments are available, ordered by relevance:

• -m or --model: Selects which type of model will be used for intent classification. Available options are: RF (uses a Random Forest), neural (uses a multilayer perceptron), keyword (uses keyword matching), and majority (always predicts the majority class). Usage: -m RF or --model RF.
• -l or --levenshtein: Allows you to specify the maximum Levenshtein distance between a user input and a keyword for it to be considered a match. Usage: -l 2 or --levenshtein 2.
• -d or --delay: Allows you to specify the delay between the user's input and the bot's response. Usage: -d 0.5 or --delay 0.5.
• --typing-speed: Allows you to speed up or slow down the typing speed of the bot. Larger numbers speed up, smaller numbers slow down. Usage: --typing-speed 0.5.
• --tts: Enable text-to-speech for bot responses.
• --speech: Enable speech recognition for user input.
• -v or --verbose: Enables verbose logging, this is meant for debugging purposes only.
• --caps: All output will be in CAPS-LOCK.

Linting and formatting

When making changes, please make sure your code is properly linted and formatted. Please use Flake8 for linting and Black (pip install black)for formatting.