This year humanity had to overcome one of the most unfortunate events in history, the COVID-19 pandemic, due to this we have had to resort to making all our communication dependent on existing technology. For this same reason we decided to make MinuteKeeper. A platform that facilitates communication through calls, providing a tool that can help many students and employees who now work from home.
What it does
MinuteKeeper is a webpage capable of keeping your meeting information neatly organized and automatically extracts important information using natural language processing. It also creates calendar events based on whether or not a due event was mentioned.
How I built it
When the user's audio dialogue is received by a python file we use Amazon Comprehend with features like keypharses, sentiment analysis, and entity analysis to get our desired information from the conversation. Finally it responds to the application with a JSON file. The application uses AWS AppSync and AWS amplify to store this conversation to the user’s profile after its authentication, we used Dynamo DB. The front-end of the web application was built using React and, Material UI and Bootstrap.
Accomplishments I’m proud of
Learning the basics of natural language processing was something really challenging and very fun. Since this was also our first time using any aws product we found it easy to use and extremely helpful.
What I learned
We learned a lot about aws products, natural language processing, and also about servers and authentication.
What’s next for MinuteKeeper
We are looking for ways to keep improving our app, to find more accessible ways for its use (we are considering a navigator extension that listens and automatically transcribes a conversation), and implementing context matching to find more useful data from the conversations.