Perform a classification task on mental issue via text
-
We built a NLP model for this classification task. We used a state-of-the-art pretrained model called
DistilBERT base uncased
, which is available onhuggingface
: https://huggingface.co/distilbert-base-uncased -
Our rationale is based on the paper "DistilBERT, a distilled version of BERT: smaller, cheaper and lighter" of SANH et el. (2020): https://arxiv.org/pdf/1910.01108.pdf
-
We also attempted to use Logistic Regression for this task, based on the paper "Classification of mental illnesses on social media using RoBERTa" by Murarka et el. (2021): https://aclanthology.org/2021.louhi-1.7.pdf
Our training data both_train.csv
is retrieved from Murarka et el. (2021) github repo:
- https://github.com/amurark/mental-health-classification
- https://drive.google.com/drive/folders/11aW_fpXjA-O51uv3xYY3xj6NWGh1VYh_
- Viet Anh Duong: Project Architecturer, NLP model maker
- Thu Ha Nguyen: Project Leader, UI designer
- Long Khanh Tran: Techinical Lead, fullstack developer
Requirements: Git, Python >= 3.7.11, npm, and node.
To replicate our results, the required dependencies are specified in the requirements.txt
file.
(Optional) A virtual machine is highly recommended. Set up guidance:
- With conda:
conda create -n venv python=3.7
conda activate venv
pip3 install -r requirements.txt
- On terminal:
npm run dev
- Open another terminal:
cd mentalist
python3 manage.py runserver
A local development server is located at
http://localhost:8000/