Framing the Israel War on Gaza: A Shared Task on News Media Narratives

Shared Task Registration Link

 Introduction and Motivation

In response to the evolving landscape of media representation and discourse surrounding the Gaza-Israel 2023-2024 war, we propose an innovative shared task aimed at delving into the intricate nuances of bias and double standards prevalent in news articles. This initiative seeks to explore diverse perspectives, cultures, and languages, fostering a comprehensive understanding of the events through the lens of major news outlets across the globe. This shared task seeks to unravel the layers of bias and propaganda within news articles in multiple languages, fostering a collaborative exploration of media narratives surrounding one of the most critical moments in recent history.

The overarching objective is to establish a shared corpus for comprehensive annotation across various layers, crafting annotation guidelines shaped by the diverse range of conflicting discourses around this sensitive topic. This endeavor seeks to bring to light both challenges and commendable aspects within the data and to foster a collaborative community. Furthermore, it aspires to nurture the growth of the next generation of Natural Language Processing (NLP) researchers, equipping them with the skills to engage directly with raw data sources.  

A different kind of Shared Task: This shared task is more like a research-oriented datathon with a strong focus on the development of improved annotation guidelines for complex opinion data tasks.  As such, it is not one that lends itself to plug-and-play ML-oriented shared tasks, but rather focuses on the more subtle aspects of data annotation and target design. That said, participants are welcome to make use of any technologies (existing or new) as part of their annotation process.  The evaluation will be done in terms of soundness of annotation methodology and guidelines,  quality, quantity, and consistency of the annotations.  All the resulting data and annotations will be shared with the community to encourage future research.

Scope of the Shared Task

Duration: The shared task is set to unfold over 8 weeks, spanning from February 29 to April 30. During this time, participants will embark on a collaborative journey to dissect news articles about pivotal moments in the Gaza-Israel war.

Multilingual and Multicultural Corpus: The cornerstone of this endeavor is a curated corpus comprising news article headlines and advertising posts about them from Facebook (henceforth, posts) in English, Arabic, Hebrew, French, and Hindi.  We collected thousands of such posts from verified accounts of news agencies from around the world in the five languages of interest mentioned above. The posts will be selected from October 1, 2023 to January 31, 2024. The total number of engagements based on total interaction will be used to rank the posts. (We will decide on exact cutoffs as we collect the data.)  The focus query will be “Gaza” in the various languages (עזה غزة गाजा and Gaza).

Key Moments of Focus: The data examines some significant key moments during the conflict, including the official war declaration, bombings at Jabaliya refugee camp and Al-Shifa Hospital, the ceasefire and hostage release, the bombings at Saint Porphyrius Church, reports on journalists and children fatalities by December, mass arrests in Gaza, and the International Court of Justice case.

Data Selection:  A subset of the full corpus will be selected for shared task annotation and it will be organized in batches of size 1,000 posts/batch (200 from each of the five languages; and 20 from each of the 10 moments).  There will be 15 such batches (total of 15,000 posts) shared. We understand that not everyone speaks the five languages, therefore we will provide machine translations into English and Arabic for all languages as applicable. We consider the machine translation to be valid stand-ins for the annotation tasks as this reflects the reality that many people have to rely on such technologies. We acknowledge that this may result in an interesting mix of human and machine biases. 

Contact

Each team member is requested to join the dedicated Google Group  https://groups.google.com/g/fignews2024/ 

If you have any specific questions about the shared task, feel free to email: wajdiz@gmail.com