The post Covid-19 situation will increase the demand for essential medicines to ensure better healthcare and personal hygiene. Finding the required medicine, without going out of home,from the nearest shop has become a preference for all. To assist this cause, the project we have build is medBOT.
What it does
medBOT helps the users to query about the nearest medicine shop for the essential medicine. It automates the query process through a chatting experience so that the user doesn't have to go through a catalog of shops to set an order. medBot recommends and performs order placement in nearest medicine shop for the user.It finds the requirements through Natural Language Processing. Also it has a suggestion module for the
How I built it
I used python to build the medBOT. Python's Natural Language Processing package 'NLTK' performs tokenization and stemming the words. Then pyTorch was used to train the model using the stems. Google Colab was used as IDE.
Challenges I ran into
pyTorch installation in local machine is a cumbersome process . Making the dataset from scratch was a tiresome work. Training on a large dataset without GPU support creates a long loading runtime issue. Hosting the server in local host has been an issue in my part so I had to end up using co-lab command based IDE for demonstration.
Accomplishments that I'm proud of
I could implement the chatBot that will work as a Brain/Backend for my system. As I was a one man team,this was a great thing for me.
What I learned
I learnt Natural Language Processing, NLTK tokenizing, building and training neural networks using pyTorch.
What's next for medBOT
Giving a nice appbased UI Recommendation of shop using Travelling Salesman Recommender system