Inspiration
This was our first hackathon, and we wanted to build something that would challenge our skills. We decided on a chatbot using Rasa, AWS, and React, aiming to provide a user-friendly interface where people could interact with a chatbot that handles natural language processing in the backend. We wanted to learn more about full-stack development, API integration, and working with cloud services.
What it does
healthbot allows users to interact with a chatbot built using Rasa which provides relevant responses to user inputted health-related queries.
How we built it
We started by setting up Rasa for the backend, configuring the intents and responses. Then, we focused on building the frontend using React, where users can type messages and receive replies. The challenge was integrating the frontend with the backend, which involved connecting the React app to the Rasa API through a REST webhook. We deployed the backend on AWS and connected the frontend through WebSocket for real-time messaging.
Challenges we ran into
One of the biggest obstacles was connecting the frontend to the backend. We ran into multiple WebSocket connection errors, CORS issues, and deployment challenges on AWS. We struggled to get the WebSocket connections to work properly, and we had to troubleshoot CORS settings to make the frontend and backend communicate. Deployment on AWS was also a challenge.
Accomplishments that we're proud of
Despite all our difficulties connecting frontend to backend, we're proud that we were able to build up a frontend and backend at all. Neither of us had much experience building websites, let alone using NLP technologies and managing AWS.
What we learned
We gained experience in troubleshooting real-time issues, integrating APIs, and deploying a system on the cloud. We also learned how to work under pressure, adapt quickly, and collaborate effectively as a team. In addition, this project gave us a helpful introduction to languages and technologies such as React, Rasa, HTML+CSS, and JavaScript.
What's next for healthbot
If we can have it deploy successfully every time, we'd like to move our focus to increasing the accuracy of the chatbot by training it more thoroughly. Beyond that, we would definitely consider adding more functionality, like personalized health advice, and expanding healthbot's knowledge base by adding more training data. We want to create a solid, usable technology that's accessible and effective.
Log in or sign up for Devpost to join the conversation.