Skip to content

The NewSHead dataset is a multi-doc headline dataset used in NHNet for training a headline summarization model.

Notifications You must be signed in to change notification settings


Folders and files

Last commit message
Last commit date

Latest commit



13 Commits

Repository files navigation


This repository contains the raw dataset used in NHNet [1] for the task of News Story Headline Generation. The code of data processing and training is available under Tensorflow Models - NHNet.

A news story is defined as a list of articles about the same event with a coherent topic. The released dataset contains 369,940 English stories with 932,571 unique URLs, among which we have 359,940 stories for training, 5,000 for validation, and 5,000 for testing, respectively. Each news story contains at least three (and up to five) articles.

The dataset is collected from news stories published between May 2018 and May 2019, where a proprietary clustering algorithm iteratively loads articles published in a time window and groups them based on content similarity1. Up to five representative articles are picked from the cluster for generating the story headline2. Curators from a crowd-sourcing platform are requested to provide a headline of up to 35 characters to describe the major information covered by the story.

Example Headlines:

  • International Space Station flyover
  • Drilling for oil in Pakistan
  • Review of 'Mr. Local'
  • MLB: Pirates vs Padres
  • Braves re-sign Jerry Blevins

Download Link

Tools to Process


If you use or discuss this dataset in your work, please cite our paper:

  title = {{Generating Representative Headlines for News Stories}},
  author = {Gu, Xiaotao and Mao, Yuning and Han, Jiawei and Liu, Jialu and Yu, Hongkun and Wu, You and Yu, Cong
and Finnie, Daniel and Zhai, Jiaqi and Zukoski, Nicholas},
  booktitle = {Proc. of the the Web Conf. 2020},
  year = {2020}


We did broad topic analysis for the 932,571 articles in our dataset. A histogram is attached as below.

Among all the 369,940 stories, each headline is required to be between 10 and 35 characters.

Such lengths of curated story headlines are much shorter than traditional summaries, and even shorter than article titles in our dataset depicted below


[1] Xiaotao Gu, Yuning Mao, Jiawei Han, Jialu Liu, Hongkun Yu, You Wu, Cong Yu, Daniel Finnie, Jiaqi Zhai and Nicholas Zukoski "Generating Representative Headlines for News Stories": World Wide Web Conf. (WWW’2020).


1 Clustering algorithm could contain noise. It is possible if some articles in a story are not relevant to the rest.

2 These articles presented don't necessarily map to articles we would show on Google products such as Search and News App.