NakbaVirality: Multimodal and Textual Virality Prediction in High-Stakes Discourse

1. Motivation and Introduction

Social media has become the primary battleground for narrative control during geopolitical conflicts. The discourse surrounding the Nakba and the post-October 7th war on Gaza represents a unique intersection of historical trauma, real-time war reporting, and highly polarized sentiment.

Understanding what makes a post “go viral” in this specific context is critical for analyzing information diffusion, propaganda spread, and public sentiment. This shared task proposes a novel challenge: predicting the reach (virality) and engagement (interaction) of posts where the content is emotionally charged, historically deep, and often multimodal (text combined with graphic or symbolic imagery).

By focusing on this specific domain, we aim to push the boundaries of how NLP and Computer Vision models handle context-heavy, sensitive, and polarizing data.

2. Data Description

We present a curated, multi-platform dataset specifically designed for this task.

• Data Sources: X (formerly Twitter) and Reddit.
• Dataset Size: 5,000 distinct samples (approx. 2,500 per platform).
• Scope: Content published post-October 7th, 2023, filtered using keywords related to “Gaza,” “Nakba,” “Palestine,” “Israel,” and specific conflict-related terminology.
• Data Fields:
  • Text: The raw post content (caption/tweet).
  • Visual: The associated image or video thumbnail.
  • Metadata: Timestamp, user follower count (for normalization).
  • Ground Truth Labels: Like count, Share/Retweet count, Comment count.

Note on Ethics: All data will be anonymized to protect user privacy, given the sensitive nature of the topic. IDs and handles will be hashed.

3. Task Definitions

We propose two distinct tasks to evaluate model performance on different modalities and prediction objectives.

Task 1: Multimodal Virality Classification

This task focuses on the interplay between text and imagery. In conflict zones, an image often determines the spread of a post more than the text.

• Input: Post Text + Post Image.
• Objective: Classify the post into virality buckets based on a normalized “impact score” (derived from likes and shares).
• Classes:
  • Low Virality: Content with minimal reach.
  • Medium Virality: Content with average engagement.
  • High Virality: “Viral” content (top 10% of the dataset).
• Evaluation Metric: Macro-F1 Score (to account for potential class imbalance).

Task 2: Textual Virality and Interaction Prediction (Regression)

This task isolates the textual component to understand how rhetoric, sentiment, and specific keywords drive engagement.

• Input: Post Text only.
• Objective: A multi-output regression task to predict two distinct values:
  • Likability ($y_{likes}$): The normalized number of likes/upvotes.
  • Interactivity ($y_{comments}$): The normalized number of comments/replies.
• Reasoning: A post may be highly liked (agreement) but have few comments, or highly commented on (controversial/polarizing) but have few likes. Distinguishing these is crucial.
• Evaluation Metric: Pearson Correlation Coefficient ($r$) and Mean Squared Error (MSE).

4. Baseline Systems

To assist participants, the organizers will provide the following baselines:

• For Task 1: A late-fusion architecture using BERT (for text) and ResNet-50 (for images), concatenated into a softmax classifier.
• For Task 2: A RoBERTa-large model fine-tuned on the regression targets with a simple linear head.

5. Significance and Impact

This shared task contributes to the NLP community by:

• Contextual Modeling: Testing how models handle “dog whistles” and specific historical references (Nakba) that require external knowledge.
• Cross-Platform Analysis: Comparing virality features between X (short-form, immediate) and Reddit (community-based, threaded).
• Multimodal Alignment: Understanding how mismatched modalities (e.g., a peaceful image with violent text) affect virality in conflict scenarios.

6. Tentative Schedule

• Jan 1: Release of Training Data (3,500 samples).
• Feb 1: Release of Development Data (500 samples).
• Feb 15: Evaluation Period Begins (Test set: 1,000 blinded samples).
• Feb 20: Evaluation Period Ends.
• Mar 1: Paper Submission Deadline.

7. Organizers

• Saad Ezzini, King Fahd University of Petroleum and Minerals
• Salima Lamsiyah, University of Luxembourg
• Shadi Abudalfa, King Fahd University of Petroleum & Minerals
• Samir El-Amrany, University of Luxembourg
• Walid Alsafadi, University College of Applied Sciences Gaza

Participation Guidelines

• For participation guidelines, please refer to Participation Guidelines.
• Comprehensive instructions for preparing and submitting your paper(s) are available at Paper Submission Guidelines.