nishi1401/detecting-minority-stress

Detecting Minority Stress

• LGBTQ+ individuals face minority stress due to stigma, prejudice, and discrimination.
• Social media, such as Reddit, is a valuable platform for detecting and classifying minority stress.
• This project explores machine learning models to detect minority stress in Reddit posts.

Methodology

• Annotated dataset used for training machine learning models.
• Self-curated dataset labeled into two classes: stress or no stress.
• Performance evaluation of deep learning and ensemble models.
• Experiments conducted with unbalanced, oversampled, and undersampled datasets.
• Different vectorization techniques used (Bag-of-Words, TF-IDF, GloVe).

Results

• Ensemble learner based on Max-Voting performed the best on an unbalanced dataset of 12,073 Reddit posts.
• TF-IDF vectorization yielded an accuracy of 85.2% and an AUC-ROC value of 0.984.

Minority Stress in LGBTQ+ Community

• LGBTQ+ individuals face challenges in coming out and often encounter rejection.
• 40% of LGBT adults have experienced rejection from family or close friends.
• LGBTQ children often experience harassment and assault at school, affecting their mental health.

Minority Stress Theory

• Stressors from a homophobic culture contribute to health disparities in sexual minorities.
• Lifetime harassment, mistreatment, discrimination, and victimization are common stressors.

Tech Stack

Python, Scikit-Learn, Imbalanced-learn, Keras, NLTK

Keywords

• LGBT
• Minority Stress
• Convolutional Neural Network
• Social Media
• Machine Learning