aansh2003/lstm-chatbot

An NLU(LSTM) chatbot which undestands simple text to generate apt responses

LSTM Based NLU chatbot with randomized responses

This is a chatbot designed to classify inputs and generate apt outputs using Natural Language Processing techniques. This project is currently active under development and will be updated from time to time.

Installing dependencies

Install pandas and numpy for dataframe and mathematical features

pip3 install pandas
pip3 install numpy

Install PyTorch for model support

pip3 install torch torchvision torchaudio

Install keras for preprocessing

pip3 install keras
pip3 install Keras-Preprocessing

Installing BeautifulSoup for HTML parsing and JamSpell for spell-check

pip3 install beautifulsoup4
pip3 install jamspell

All required dependencies have now been installed.

Additional dependencies for flask server deployment

pip3 install Flask

Running the model from command line

Type python3 out.py and wait a couple of seconds till it shows

Started...
user:

Here is an example output:

Running the model using a flask server

Type python3 flask_test.py and wait till the server has been deployed. A link will be displayed like the following

Running on http://127.0.0.1:5000

Copy the link at paste it on your browser to deploy the website. Here is an example output:

Currently working on a better deployment method for the model with a proper front-end and also further increasing the number of classes and accuracy of the model and also adding numerous other responses.