Emotion.Net - Computers have feelings too!
Visit our live demo here!
People matter. Emotions matter. The COVID-19 pandemic has put people's mental health at risk. Research has shown that the pandemic has nearly tripled the rate of depression. We created EmotionNet to be able to assess the feelings of individuals during hard times.
EmotionNet asks users to type sentences and words which describe their day or current state of mind. Based on this description, it scores the user’s positivity level on a scale of
-100% (being the most negative) to
100% (being the most positive). The interface updates in real-time, providing a responsive experience. Based on that, EmotionNet can give the user a joke to cheer them up if they are extremely negative.
How we built it
Challenges we faced
While making EmotionNet, we encountered numerous challenges. Here is an unordered list of some obstacles we met:
- Since this is the first time the team attended a hackathon, we were not accustomed to the limited time we had!
- The Azure Cognitive Service had a strict 5000 requests per month limit, but since we wanted to update in real-time, we had to improvise. Our team came up with a compound text analyzer that combined a local instance of NLTK which offered real-time performance, with the cloud to offer the most accurate predictions.
- We had limited knowledge in web-development, and had issues properly aligning items when the page became complex.
What we learned
Building EmotionNet was an excellent team-building experience. Through the limited time we had, we not only learned more about developing websites and services, but more importantly, we learned more about each other!
Opportunities for expansion
EmotionNet has plenty of potential for further development. We hope to increase the accuracy of our prediction software to provide for a better experience with our users. Other potential opportunities include adding multiple emotions, an emotion detection API, and improved performance.