Many people engage in self-reflection through journaling or diary writing, yet they find it difficult to interpret how their words connect to their emotions.
We wanted to work on a product to improve emotional wellbeing for social good.
What it does
Reads each sentence of your journal
Predicts the emotion behind it
Combines sentence-level predictions into daily emotional summaries
How we built it
Each sentence is analyzed with a pretrained BERT model
The model predicts one of 20+ emotion categories
These are aggregated across entries to show moods, analyze the reasons for the predicted mood and suggest ways to improve mental health.
Challenges we ran into
Analyzing models for better performance for emotional data.
Building a visually appealing UI since the UI should also designed to destress the users.
Algorithm for emotional detection for data with long paragraphs.
Accomplishments that we're proud of
Creating a web app
Predictions with visualizations for analysis.
What we learned
We learned that to create a project we should manage time more efficiently by setting deadlines.
What's next for MoodMaze
We will be porting to mobile application that will easily import multi-modal data for easier logging of your emotions.
Log in or sign up for Devpost to join the conversation.