We wanted to create an application that people could use when they are feeling depressed, angry, or other negative emotions.
What it does
It uses machine learning to analyze the emotions of statements made by the user, and using the predicted emotion it tries to find a suitable Spotify playlist that the user can listen to so that they can feel better.
How we built it
We built the front-end of the project using basic HTML, CSS, and JS developed and hosted on QOOM. To build the Machine Learning model we used PyTorch and trained it on google-colab.
Challenges we ran into
One of the challenges we ran into is that the first dataset we used was really bad, and had bad examples. We decided to use a different dataset proposed at an ML conference, and that one worked a lot better since the tweets within the data had better language-sentiment mappings. We also found it difficult to try and push the backend code onto google cloud and were unable to actually get the backend onto google cloud so you have to host the backend locally.
Accomplishments that we are proud of
We are proud that we got to finish the project and get a decently working ML model, as well as making a frontend that looks nice.
What we learned
We learned how to use QOOM, and we also learned how to push a basic app onto google cloud.
What's next for Meowtalk
Adding a tracking service to show how your emotions are changing over multiple days. Also adding more emotions to the Machine Learning model