Why Are We Doing This?

In today’s digitally connected world, online polarization is rapidly rising—splitting communities, amplifying misinformation, and worsening conflicts. Social media and online discourse are filled with increasingly hostile language targeting political views, religions, ethnic groups, and even entire nations.

While NLP has made impressive strides in sentiment analysis and hate speech detection, polarization remains understudied, especially across languages, cultures, and real-world events.

What Do We Want to Achieve?

This task sets out to develop and evaluate NLP systems that can:

• Detect polarization in diverse texts and languages
• Understand who or what is being targeted—religions, political ideologies, races, or genders
• Interpret how polarization is expressed—via framing, dehumanization, stereotyping, and more
• Operate effectively across high-, mid-, and low-resource languages
• Work on real-world events like elections, protests, and international conflicts
• Provide interpretable outputs, not just predictions
• Promote ethical, explainable AI in socially sensitive contexts

Long-Term Vision

By building robust, explainable models of online polarization, we hope to:

• Enable early detection of toxic or polarizing discourse
• Support moderation and peacebuilding efforts
• Advance cross-linguistic and cross-cultural NLP
• Facilitate interdisciplinary research at the intersection of language, society, and technology
• Build datasets and benchmarks that can be reused and extended by the community

A Step Toward Responsible NLP

We believe NLP should not only understand what people say, but also why they say it and how it impacts society. This task contributes to that goal by creating:

• Meaningful annotations
• Interpretable models
• Public datasets
• Open discussions about bias, ethics, and fairness

Your Contribution Matters

Every team that participates helps push the boundaries of multilingual, socially-aware AI. Together, we can build systems that do more than classify—they can help explain, caution, and connect.

Join us in making NLP not just more powerful, but also more responsible.