Corroborated Anonymous feedback system

Telling your friends hard truths is tough and listening to them might be even tougher. Thankfully, technology can let us avoid the awkwardness while still experiencing the growth from getting feedback

Comment

Inspiration

I've always found it hard both to give and receive feedback from friends. When on the receiving end, getting feedback when I'm not ready to hear it can make me feel defensive which makes me less receptive. On the flip side, giving feedback is hard because I do not know if the feedback I want to give is just a personal issue I have with the person or something that is truly a large problem in their life. One possible solution is to discuss it with the persons's other friends before bringing it up to him but that can easily be misconstrued as talking behind the person's back which would make them feel even more defensive.

What it does

The app allows for anonymous yet corroborated feedback through the use of semantic similarity analysis. This allows people to give others feedback without having to worry about whether or not their feedback is significant since your feedback will only be shown to the person if enough people corroborate it. From the receiving end, this allows a person to set the number of corroborated instances of feedback before they are notified of it. This also allows the user to open up the feedback when they feel they are in the best mindset to be receptive to it.

How we built it

We used open source frameworks for semantic similarity analysis and fine tuned it towards our use case. We then used Python GUI to create visualize how the corroborative part of the application works

Challenges we ran into

We had no experience with Natural Language Processing and we had a tough time getting the model to work

Accomplishments that we're proud of

Not giving up and falling back on things we are more technically familiar with

What we learned

We learnt about the various ways meanings behind sentences can be extracted such as vectorization and we also got a look at the tremendous amount of math behind it.

What's next for Corroborated Anonymous feedback system

Currently we only have the corroborative model working. Going forward, we would need to create the rest of the application to make it deploy-able. We would also need to fine tune the semantic similarity analysis further

Built With

Share this project:

Updates