NLP-Based Document Classifier

This project provides tools for unsupervised document clustering using NLP based approaches. It includes:

• st_workflow: Sentence transformer based document clustering
• d2v_workflow: Document2Vec-based document clustering

Getting Started

1. Install Dependencies

We use uv as a faster and more reliable alternative to pip.

Install uv:

pip install uv
uv sync

2. Run Workflows

Standard Clustering Workflow

uv run ./st_workflow/st_main.py -i documents/

• -i: Path to a folder containing your text documents.

• -o: (Optional) Output folder. Defaults to st_clusters/.

Doc2Vec Clustering Workflow

uv run ./d2v_workflow/d2v_document_clustering.py

Notes

• st_workflow uses a pre-trained model & supports multiple formats: txt, pdf, docx, html, htm.
• d2v_workflow was trained using the ag_news dataset.