Many popular news sources show bias in their news headlines. We decided to create a project that would help educate people about bias in news by showing how different news sources have reacted to COVID-19.
What it does
Our website uses sentiment analysis techniques to analyze various news sources to see how they reacted to the pandemic based on headlines of articles they publish.
How I built it
We used Node.js for the frontend of the website, and we used an extractive algorithm for semantic analysis. However, we originally developed a transfer learning based approach but were unable to implement it with the frontend due to time constraints.
Challenges I ran into
We did not have enough time to implement all of the features we would have liked to, and the neural network took around 4 hour to train, preventing us from experimenting with it more than we would have liked to. Additionally, it was our first time using Node.js, which was a learning experience as well.
Accomplishments that I'm proud of
I am proud that were able to finish the neural network based model, and that we were able to create such a good website even though it was our first time using Node.js.
What I learned
We learned how to use Node.js, and work in a more time-efficient manner to create a complete product. Additionally, we learned how to better use transfer learning techniques.
What's next for Sentiment Analysis of COVID-19 related Headlines
In the future, we will advance our project by making it able to update rss feeds, and show how the data has changed over time, also comparing it to actual COVID cases. Additionally, we would like to analyze entire articles instead of just news headlines, and move towards a more abstractive, deep-learning based approach.