walzen/gigaword

---
license: apache-2.0
task_categories:
- summarization
language:
- en
size_categories:
- 1M<n<10M
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
---

# Gigaword (Repackaged)

Putting it here for better accessibility.

## Description

Headline-generation on a corpus of article pairs from Gigaword consisting of around 4 million articles. Use the 'org_data' provided by https://github.com/microsoft/unilm/ which is identical to https://github.com/harvardnlp/sent-summary but with better format.

There are two features:
- **document**: article.
- **summary**: headline.

**Homepage:** https://github.com/harvardnlp/sent-summary

**Original Source code:** tfds.summarization.Gigaword

## Dataset Statistics

- **Versions:** 1.2.0 (default)
- **Download size:** 551.61 MiB
- **Dataset size:** 1.02 GiB

### Splits

| Split | Examples |
| :--- | :--- |
| 'test' | 1,951 |
| 'train' | 3,803,957 |
| 'validation' | 189,651 |

## Feature Structure

```python
FeaturesDict({{
'document': Text(shape=(), dtype=string),
'summary': Text(shape=(), dtype=string),
}})
```
Citation

```
@article{{graff2003english,
title={{English gigaword}},
author={{Graff, David and Kong, Junbo and Chen, Ke and Maeda, Kazuaki}},
journal={{Linguistic Data Consortium, Philadelphia}},
volume={{4}},
number={{1}},
pages={{34}},
year={{2003}}
}}

@article{{Rush_2015,
title={{A Neural Attention Model for Abstractive Sentence Summarization}},
url={{[http://dx.doi.org/10.18653/v1/D15-1044](http://dx.doi.org/10.18653/v1/D15-1044)}},
DOI={{10.18653/v1/d15-1044}},
journal={{Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing}},
publisher={{Association for Computational Linguistics}},
author={{Rush, Alexander M. and Chopra, Sumit and Weston, Jason}},
year={{2015}}
}}
Original TFDS Catalog: https://www.tensorflow.org/datasets/catalog/gigaword