README.md

---
title: Lesbian Greek Morphosyntactic Parser
emoji: 🔍
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.35.0
app_file: app.py
pinned: false
license: cc-by-4.0
---

# Lesbian Greek Morphosyntactic Parser

A Hugging Face Space for parsing dialectal Greek text from the island of Lesbos using the Lesbian Greek Morphosyntactic Model developed by Bompolas et al. (2025).

## Overview

This interactive parser provides morphosyntactic analysis for the Lesbian dialect of Greek, offering:

- **Part-of-speech tagging**
- **Morphological analysis** 
- **Dependency parsing**
- **Lemmatization**
- **CoNLL-U format output**

## Model Details

The underlying model is based on:
- **Stanza v1.7.0+** as the base pipeline
- **Greek BERT** ([nlpaueb/bert-base-greek-uncased-v1](https://huggingface.co/nlpaueb/bert-base-greek-uncased-v1)) for enhanced representations
- **UD_Greek-Lesbian treebank** for training (540 sentences)

### Training Data Sources

**Oral Data** (collected 2023-2024):
- Speakers from Agra, Chidira, Eressos, Pterounta, Mesotopos, and Parakoila villages on Lesbos

**Written Sources**:
- Papanis, D. & Papanis, G. D. (2004). *Lexiko tou Agiasotikou Glosikou Idiomatos*
- Tsokarou-Mitsioni, E. (1998). *Palies Istories ap' tn Agiasiou*
- Tsokarou-Mitsioni, E. (2019). *Prosfygiá*
- Anagnostopoulou, M. A. (2021). *Thematiko Lexiko tis Lesviakis Dialektou*
- Anagnostou, V. T. (2014). *Tsi sta th'ka mas: Komodia sta k'stariot'ka*

## Features

### 📊 **CoNLL-U Output**
Standard Universal Dependencies format for interoperability with linguistic tools

### 📈 **Interactive Data Table**
Browse parsed tokens with all linguistic features (POS, morphology, dependencies)

### 🔗 **Dependency Visualization**
Text-based visualization showing syntactic relationships between words

### 🏛️ **Dialectal Specialization**
Optimized specifically for the Lesbian dialect of Greek

## Usage

1. Enter your Lesbian Greek text in the input field
2. Click "Parse Lesbian Greek Text" or press Enter
3. View results in three formats:
   - Raw CoNLL-U output (copyable)
   - Interactive data table
   - Dependency structure visualization

## Example Texts

The interface includes example texts based on the dialectal sources used in training:

- `Το παιδί κάθεται στο σπίτι.`
- `Η μάνα μαγειρεύει στην κουζίνα.`
- `Το νερό τρέχει απ' τη βρύση.`
- `Οι παππούδες λένε παλιές ιστορίες.`

## Limitations

- **Experimental model**: Due to limited training data (540 sentences)
- **Domain-specific**: Optimized for dialectal content similar to training sources
- **Research purposes**: Further fine-tuning needed for production use

## Citation

If you use this tool or the underlying model, please cite:

```bibtex
@inproceedings{bompolas2025crossing,
    title={Crossing Dialectal Boundaries: Building a Treebank for the Dialect of Lesbos through Knowledge Transfer from Standard Modern Greek},
    author={Bompolas, Stavros and Markantonatou, Stella and Ralli, Angela and Anastasopoulos, Antonios},
    booktitle={Proceedings of the 8th Universal Dependencies Workshop (UDW, SyntaxFest 2025)},
    year={2025},
    publisher={Association for Computational Linguistics}
}
```

## Related Resources

- 🤗 [Lesbian Greek Morphosyntactic Model](https://huggingface.co/sbompolas/Lesbian-Greek-Morphosyntactic-Model)
- 📚 [UD_Greek-Lesbian Treebank](https://github.com/UniversalDependencies/UD_Greek-Lesbian)
- 🔧 [Stanza Documentation](https://stanfordnlp.github.io/stanza/)
- 🇬🇷 [Greek BERT Model](https://huggingface.co/nlpaueb/bert-base-greek-uncased-v1)

## Technical Details

### Dependencies
- `gradio>=4.0.0` - Web interface
- `stanza>=1.7.0` - NLP pipeline
- `pandas>=1.5.0` - Data handling
- `torch>=1.9.0` - Neural network backend
- `transformers>=4.20.0` - BERT integration

### File Structure
```
├── app.py              # Main Gradio application
├── requirements.txt    # Python dependencies
└── README.md          # This documentation
```

## Development

To run locally:

```bash
git clone <this-space>
cd <space-directory>
pip install -r requirements.txt
python app.py
```

## Support

For issues related to:
- **The model**: Contact the original authors or open an issue on the model repository
- **This Space**: Open an issue in the Space's discussion tab
- **Stanza**: Refer to the [Stanza documentation](https://stanfordnlp.github.io/stanza/)

## License

Please refer to the original model's license terms and the individual component licenses (Stanza, Greek BERT, etc.).