A chatbot for Pizza ordering service, as per user customization, supporting queries and the responses in more than one language.

Conventional Chatbots though are very useful, lack support for local languages. The goal of this project is to overcome that disadvantage and create a multilingual chatbot supporting at least 2-3 languages.

.
├── Readme.md
├── Results.pdf
├── chatgui.py
├── models
│   ├── chatbot_model.h5
│   └── lid.176.ftz
├── requirements.txt
├── resources
│   ├── classes.pkl
│   ├── intents.json
│   ├── pizza.png
│   └── vocab.pkl
├── train_chatbot.py
└── utils
    ├── api_utils.py
    ├── chatbot_utils.py
    ├── fasttext_utils.py
    ├── transformer_utils.py
    └── translation_utils.py

Clone this repository from GitHub and open the root directory in the terminal.

The required pip packages for successfully executing the project are:

numpy==1.20.1
wget==3.2
requests==2.25.1
googletrans==3.1.0a0
fasttext==0.9.2
nltk==3.6.1
Keras==2.4.3
transformers==4.5.1
Pillow==8.2.0

python train_chatbot.py 

The intents are parsed from intents.json to generate the chatbot vocabulary and train a Keras sequential model for 200 epochs.

python chatgui.py

The GUI window for the chatbot is created.

Each user input and chatbot response (with source language and translation) are printed in the terminal too.

Results for execution of the Multilingual Chatbot for Pizza Ordering can be found in this file.

The results file shows chatbot conversations for 4 case scenarios along with screenshot of GUI.