SpeechFire

Login
Main Page
Post and Reply
Profile

Inspiration

As teenagers, we felt that social media has caused stress in our and our friends' lives. We also consider ourselves innovators, and we believe that there currently isn't any platform built from the ground up for idea sharing. On many platforms, innovators sharing their idea would be ridiculed.

What it does

We believe we have created a solution to these issues in our application. The first is addressing the ridicule. We use a sentient analysis AI to identify if people posting comments and ideas are positive. If it's not, then we prompt the user if they're sure if they want to post it. We believe this will largely change the behavior across the platform since people would be forced to go out of their way to post negative comments, while positive comments are easier to write.

Additionally, we encourage users to post longer, technical articles, about their ideas or positions by having a minimum post length of 100 characters. This is directly opposed to platforms like Twitter which has a maximum length, which causes users to condense their ideas and construct straw-man arguments.

How we built it

We created the app using Flask, Vue, and Tailwind, which integrates with SQL. The SQL contains two tables, users, and posts, which contain information vital for the app to work. The AI is built off a python module called Text Blob, which uses Twitter posts as training.

Challenges we ran into

Tailwind CSS was a learning curve. This was the first time any of us have used Tailwind, so some things which we expected to work, didn't. For example, for nearly the entire project, we couldn't get the sidebar to stretch perfectly all the way down. Despite following the documentation, nothing we worked tried. Only through a lot of trial and error, did we get the right combination of classes that worked.

Accomplishments that we're proud of

This was the most professional app we've ever made. The combination of frontend technology and the complex backend was at a scale none of us anticipated.

What we learned

Tailwind CSS was a struggle, and so was the AI. Initially, we tried to make an RNN but several libraries such as TensorFlow couldn't be imported despite our best efforts.