Doing data-driven discovery shows a lot about a person's social media account. With Twitter being one of the most used channels of communication we thought it would be awesome to see how Twitter accounts face up to these data analysis methods. Also check out Obscurify

What it does

Search a public popular twitter account and you will see the kind of overall emotion a twitter account has based on their recent tweets.

How we built it

We used the following:

  • Flask
  • React-Native Web
  • MongoDB
  • Heroku
  • Netlify

Challenges we ran into

  • Multiple calls spiked the usage rate of our sentiment analysis rates for Azure
  • Data sanitization which often had a lot of unnecessary links and text from tweets that were simply just a picture.
    • However, it would be nice to run a sentiment analysis on those pictures or videos that are sent too

Accomplishments that we're proud of

  • Sleek and simple UI design that allows for ease of use
  • Optimizations in requests allowed for more data to be analyzed as well
  • Scalability is possible with how our project is set up so that new ways of analysis can be added too

What we learned

  • A lot of unfamiliar tech stacks and APIs like Flask, MongoDB, React-Native, Text Analytics API, Twitter API
  • Open source workflow and guidelines

What's next for Twanalyze

  • Optimizations in requests in order to avoid the need of doing multiple calls to text analytics API. By accomplishing this we will be able to do more interesting NLP-driven usages and graph them to public and searchable accounts.
  • Expand the kind of data that we are visualizing here:
    • Fact-checking in a newsfeed
    • Network graphs of a follower base
    • Detect the political alignment in newsfeed based on sources that one follows
Share this project: