Howdy Vibe Check

See how TAMU is doing in the public eye on twitter, without all the reading.

Comment

Inspiration

We wanted to create an easy way for people to see the public opinion on TAMU with certain topics, for example, how the university is handling covid-19, using tweets. We wanted something that could gather tons of tweets and get an overall consensus on the sentiment of the public towards a particular topic at TAMU.

What it does

Crawls through tweets concerning TAMU in a given month and uses machine learning to divide tweets into different categories, such as covid-19, zoom classrooms, new students or freshmen, etc. It then determines the sentiment of each tweet and gives an overall rating on a topic. In short, it tells the user how the public feels about A&M in relation to an event. Is A&M handling the transition to online learning well? How does everyone feel about the pandemic situation on campus?

How I built it

We loaded in the tweets from the entire month of august, removed stopwords, lemmatized the data to remove excess words, and made an lda model using gensim to give us a general idea of different topics. We then found the polarity of each word and used that to get an overall positive/negative rating for the tweets. We used chartjs to represent our data on the website.

Challenges I ran into

There was not enough data for us to analyze. Coming up with keywords as well proved to be difficult as certain words would appear in many tweets and create topics that were too similar. The majority of us were new to web development as well, but it was cool to learn the process.

Accomplishments that I'm proud of

Able to use machine learning to determine the sentiment of tweets.

What I learned

How to incorporate our data into the website in a chart. We also learned the basics behind building a website.

What's next for Howdy Vibe Check

Allow the user to enter in specific days and be able to track how a certain topic fared over a period of a month. Add the option for the user to expand their search to multiple universities across the nation. We would also like to get the website to work with real time data so that as tweets come in we can continuously update.

Built With

Share this project:

Updates