---
license: cc-by-sa-4.0
---

# Dataset
## Topic-exclusive split
The training dataset contains popular topics (instances), while other topics are in the test and validation datasets.
* test_diff.json
* training_diff.json
* validation_diff.json

## Topic-independent split
Topics are randomly selected in datasets. For a common purpose, we suggest THESE DATASETS.
* test_random.json
* training_random.json
* validation_random.json

# GitHub
* https://github.com/declare-lab/WikiDes/ 


# Citation

## APA
Ta, H. T., Rahman, A. B. S., Majumder, N., Hussain, A., Najjar, L., Howard, N., ... & Gelbukh, A. (2022). WikiDes: A Wikipedia-based dataset for generating short descriptions from paragraphs. *Information Fusion*.

## BibTeX
```
@article{Ta_2022,	
doi = {10.1016/j.inffus.2022.09.022},	
url = {https://doi.org/10.1016%2Fj.inffus.2022.09.022},	
year = 2022,	
month = {sep},	
publisher = {Elsevier {BV}},	
author = {Hoang Thang Ta and Abu Bakar Siddiqur Rahman and Navonil Majumder and Amir Hussain and Lotfollah Najjar and Newton Howard and Soujanya Poria and Alexander Gelbukh},	
title = {{WikiDes}: A Wikipedia-based dataset for generating short descriptions from paragraphs},	
journal = {Information Fusion}}
```

# Paper links
* https://doi.org/10.1016%2Fj.inffus.2022.09.022
* https://arxiv.org/abs/2209.13101

# Contact
Hoang Thang Ta, tahoangthang@gmail.com