We live in the world of polarized opinions and ever-changing social and political milieu. We wanted to make it easier to find sources of political and social divide by monitoring the most controversial topics of the most active social network Twitter.
What it does
This service analyzes the Trending topics of Twitter and assesses the sentiment, that is, what the people's attitude to them is. Then we provide a visualization of how split the opinions are, and how well sides are balanced.
How we built it
By trial-and-error, driven by the passion for creating visualizations for the data which is too abundant or changes too quickly to be monitored in its pure form.
Challenges we ran into
Combining different skills (back- and front-end), hunger and lack of sleep, deployment.
Accomplishments that we're proud of
Visualizations that really make sense.
What we learned
How to structure the information and how to make the most of the Bokeh visualization library.
What's next for SentiTwitter
It would be great to switch from Microsoft Cognitive Services to custom machine-learning models for sentiment analysis to be able to support more languages (Finnish, Russian etc).