Since the COVID vaccines are widely available in the U.S., both national and state governments have issued very creative yet controversial measures that encourage us to get vaccinated. Curious about the public acceptance of vaccine incentives, our group decided to mine and analyze the Twitter feeds on this topic. The results show that the effectiveness of such campaign varies by state.
Particularly, New York City (free burgers), California (cash prizes), and New Jersey (free alcohol drinks) feel more positive about their vaccine incentives than Ohio (1 million lottery) and West Virginia (guns).
What it does
scraped the Tweets that reply to official announcements on vaccine rewards
feed cleaned texts into a Natural Language Processing tool for sentiment and opinion analysis as well as taxonomy classification (behavior and emotional)
visually displayed the sentiment score distribution, unique emotional traits with their frequencies, and behavior traits with their frequencies
the user can look up these results by state name. note: we only included a few most characteristics states, not all states are available for scraping because there are simply not enough tweets on the topic. For example, even though TX and FL have high vaccination rates, they do not have any special incentives.
How we built it
Web scraping: selenium and chrome driver
Natural Language Processing: Expert.ai client package in Python, cloud version (see here)
Web application: flask
Challenges we ran into
display a word cloud for each state
change visual output based on drop down menu selection
install chrome driver on different computers
Accomplishments that we're proud of
successfully scraped Tweet replies from Twitter
integrate natural language processing workflow with a web application
generate decent visuals on the website
What's next for Vaccine Sentiment
Incorporate all states on the website.