research

what we're working on

moodlog is built around a fine-tuned roberta model that detects emotions from short journal entries using plutchik's eight primary emotions. we're actively exploring:

• · emotion prediction accuracy from short-form text
• · longitudinal emotion pattern detection
• · entity-emotion correlation analysis
• · privacy-preserving approaches to mental health data

collaborate

if you're working in psychology, nlp, mental health tech, or anything adjacent - we're open to conversations. we're especially interested in hearing from researchers who want to work with emotion-labeled journal data (anonymized, opt-in only).