# Epstein Semantic Explorer — HuggingFace Edition

A lightweight, dependency-free semantic analysis tool for the
recently released House Oversight Epstein documents.

### 🚀 Features
- Cluster browser  
- BM25 semantic search  
- Topic terms  
- Entity frequency explorer  
- Cross-cluster entity mapping  
- Works entirely on CPU  

### 📁 How to Use
1. Upload your dataset: `epstein_semantic.jsonl`  
2. Select a cluster  
3. Search keywords  
4. Explore entities / topics  

### 📦 Dataset Source
Full processed dataset is available on Kaggle:  
https://www.kaggle.com/datasets/cjc0013/epstein-bge-large-hdbscan-bm25/data

### 📝 Notes
- No external API calls  
- No model downloads  
- 100% local execution inside the Space  

MIT License © cjc0013