TweetMood

@mileycyrus Twitter Vibe
@perplexity_ai Twitter Vibe
@scotiabank Twitter Vibe
@CBCNews Twitter Vibe

Inspiration

Our inspiration for this project was the Hedonometer that collected daily tweets from Twitter before it became X to summarize the happiness factor of certain days. It would use sentiment analysis to identify if the keywords in a tweet were positive, negative, or neutral. Based on scoring the "happiness" of that day would be identified. [link]https://hedonometer.org/timeseries/en_all/?from=2021-12-30&to=2023-06-29

What it does

TweetMood is a web-based application that allows you to see the positivity of a Twitter user's profile by taking random tweets and conducting sentiment analysis to categorize posts as "positive", "negative", and "neutral". Please note, this app is for educational purposes only.

How we built it

It was built using Python, PlayWright for collecting data, and Vader Sentiment for sentiment analysis, and HTML, CSS, and JavaScript for the front-end of the application.

Challenges we ran into

Our team ran into two challenges. The first challenge was with our original idea of an Outlook add-in that would use AI to summarize the urgency and importance of an email. Additionally, the add-in would allow a user to write an email and have AI to re-write it with more / less urgency and different tones. The issue was that Outlook requires all API calls to be authenticated, which includes reading or writing emails. We were able to gather the filtered bodies of emails and save them into JavaScript variables, but using the data after that was limiting. Due to these limitations, we pivoted to TweetMood.

The second challenge we ran into was that the original AI model used for sentiment analysis was from Selenium. The AI model worked well but would have blackout periods of getting requests from the internet. These blackouts would last around 30 minutes to over an hour where our team wasn't able to analyze any data from Twitter, and therefore we weren't able to test the app. To overcome this limitation and the concern of having a blackout period during the final stages of the project, our team switched to using Vader Sentiment and PlayWright.

Accomplishments that we're proud of

We are proud of the success we've had with pivoting ideas over half-way through the hackathon while still being able to submit a project for judging.

What we learned

Our team learned how Outlook uses authentication certificates to validate requests to the Outlook API and how to create custom add-ins for email apps and browser sites. We also got to learn more about how to use GIT as a team and how to deploy full applications using Versel (front-end) and Render (back-end).