SML - Identify glitchy AI-generated text
2000 Zindi Points
Help filter out bad auto-generated summaries of news articles
66 data scientists enrolled
MediaClassificationNLP
Africa
20 September 09:00—21 September 23:59
Ends in 27 hours

Our very own Jonathan Whitaker is on a mission to make it easier to keep up with the news. Part of that project involves generating summaries of news articles for easy digestion. Only one problem: these summaries aren’t always perfect!

The goal of this competition will be to build a model to weed out these low-quality summaries to improve the user experience. One other problem: We need to label the summaries!

Data collection for this competition is currently ongoing. And we are asking you - the Zindi community - to help us get this competition launched as soon as possible by labelling some summaries at https://sml-label.onrender.com/. A summary is presented (along with the article title and a link), and volunteers will flag those that are flawed in some way.

Our goal is to reach 2,000 labelled summaries by 30 August 2020! As soon as we meet our goal, we will launch this labelled dataset as a machine learning NLP competition. So watch this space!