Intent Recognition with Pytorch,Torchtext and Streamlit

Business Perspective

Everyday people use many kinds of personal assistance systems such as Google Assistant,Siri,Alexa,etc.. in those systems people usually ask questions such as "What is the weather in Bay Area!?","Add Wonder to mendes playlist",etc...,these type of questions/commands will help in increasing the training data of those systems, and these data can be grouped into common categories such as GetWeather,AddToPlaylist,etc.. so in this project a Machine learning system is created to classify these intents of various users

Description

This motivation of the project is to classify the various intents of the users into seven categories using methods of Deep Learning, I have used various architectures such as

1. RNN
2. Bidirectional LSTM
3. Fastext
4. TEXTCNN
5. BERT

Installation

If you want to test this on your own machine I would recommend you run it in a virtual environment or use Docker, as not to affect the rest of your files.

Python venv

In Python3 you can set up a virtual environment with

python3 -m venv /path/to/new/virtual/environment

Or by installing virtualenv with pip by doing

pip3 install virtualenv

Then creating the environment with

virtualenv venv

and finally activating it with

source venv/bin/activate

Pretrained weights

The pretrained weights of this project can be found here,which also contains the saved vocab file

References

These repositiries highly helped me to build this project bentrevett/sentiment-analysis