Haccolade

Inspiration

Back in high school, some 4-5 years ago, I was wondering about an innovative way to help people. It was November 5th when I came up with my little solution that came with a large impact - Bluevale Compliments. It was actually inspired by an initiative at Queens and WLU.

What it does

The account is an anonymous way for students to send compliments about their fellow peers. Over 2 years, we got about 2000 - 3000 messages posted for students at our school. That is a big number, and vetting the compliments and posting to the page is no easy task. Trust me! So, what Haccolade does is automate the process by evaluating the emotional tone of the messages sent to a facebook page. If the messages are emotionally sound, it is assumed that the message is appropriate enough to be posted.

How I built it

The project is essentially an express server deployed on Heroku. Unfortunately, AWS was taking too long to confirm. Anyways, our project integrates wit the Facebook api to receive messages and post to the designated page with user access tokens. When a message is received, our bot will implement IBM Watson to evaluate the emotional criteria (joy, anger, sadness, disgust, and fear).

Challenges I ran into

Unfortunately this solution is very early stage. There are definitely a multitude of ways in which this app can be further improved - specifically, I wanted to implement IBM Watson's natural language classifier to further determine the nature of the message (i.e is the message a question? What does it entail specifically?). However, due to problems with facebook's api - and it's documentation, my team ran into some major problems that slowed us down.