Inspiration

We were inspired to promote #HealthyConversations by showing users what the impact of their tweets are by analyzing their sentiment value.

What it does

This Google Chrome extension gives you a report of the health and credibility score of your tweet before you send it. Your tweet will turn dark red, light red, light blue, light green, or bright green depending on its score.

How we built it

Using Python, Java, Html, Flask, and CSS we were able to develop separate software that once united would make a Google Chrome extension to encourage consideration before posting and suggestions to encourage fact-checking and credibility. For the credibility algorithm, we utilized pre-existing algorithms such as subjectivity and polarity software which would return a percentage for each category. In addition, we also cross referenced the user's tweet with a database of hate speech in order to determine if the user would be flagged for inappropriate or discriminatory use. Using an iterative summation and adding it to a cushion baseline, these categories were utilized to draft the credibility score.

Challenges we ran into

Twitter APIs did not give as much user information as we would have liked. We really wanted to check if the user's comments had been flagged or if there were any indicators of offensive behavior in order to further help the user. Without knowing if they had been flagged, it was a little difficult to make specific suggestions and encourage them to be considerate on this platform.

Accomplishments that we're proud of

It works! We are so proud that our extension actually works and we hope to someday see this implemented within Twitter itself!

What we learned

It is difficult to define abstract ideas! It was very difficult to define credibility or sentiment analysis because emotions and ideas are not easily translated into workable numbers.

What's next for Happy Bird for Twitter

Further develop our extension so more parameters can be added to the credibility score. This would allow for a more accurate model and would give more insight into navigating a growing world of misinformation.

Built With

Share this project:

Updates