Sequence-to-sequence quotations attribution

This repository contains the source code for the paper "Fine-grained quotation detection and attribution in German news articles" by Fynn Petersen-Frey & Chris Biemann published at KONVENS 2024. Implementation based on Seq2seqCoref.

Setup

mamba env create -f env.yml --solver=libmamba
mamba activate seq2seq-quotations

Dataset

Available at https://github.com/uhh-lt/german-news-quotation-attribution-2024

Pre-trained models

• mT5-large

Inference / Prediction

1. Download and untar a model (see Pre-trained models)
2. Place plain-text files to process in a folder
3. Create an empty directory for temporary files
4. Create an empty directory for final output
5. Activate the correct python environment (see setup)
6. Execute below command:

python predict.py <input-dir> <temp-processing-dir> <model-dir> <output-dir>