Cohesion-6K: An Arabic Dataset for Analyzing Social Cohesion and Conflict in Online Discourse

2026-05-21 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Social Sciences & Behavioral Studies · Depth: Expert, quick

Summary

Cohesion-6K is a newly introduced Arabic dataset designed to analyze social cohesion and conflict within online discourse. Comprising 6,000 manually and ChatGPT-assisted annotated public Facebook posts related to the Israeli Occupation of Palestine, the dataset categorizes each post into one of five discourse types: Conflict, Resolution, Community Engagement, Supportive Interactions, and Shared Values. The annotation process, which combined expert human judgment with model-assisted pre-labeling, achieved substantial inter-annotator agreement with a Cohen's kappa of 0.85. Quantitative analysis of Cohesion-6K reveals a significant engagement gap, showing that conflict-oriented posts attract between two and four times more user interaction than resolution-oriented ones (p < 0.01). This resource, including its annotation guidelines and preprocessing code, will be openly released to support research in computational social science, digital communication, and Arabic natural language processing.

Key takeaway

For Research Scientists and NLP Engineers analyzing online discourse, particularly in Arabic contexts, Cohesion-6K offers a critical resource. You should integrate this dataset to develop models that accurately distinguish between conflict and cohesion narratives. The finding that conflict posts receive 2-4 times more engagement suggests your models must account for this inherent bias, preventing skewed interpretations of public sentiment. Utilize the open-licensed dataset and annotation guidelines to refine your analytical frameworks.

Key insights

Divisive online discourse consistently attracts disproportionately higher user engagement compared to unifying narratives.

Principles

• Online discourse analysis requires nuanced categories beyond toxicity.
• Conflict-oriented content receives 2-4x more user interaction.

Method

A hybrid annotation approach combines expert human judgment with ChatGPT-assisted pre-labeling, followed by trained annotator verification, achieving Cohen's kappa = 0.85.

In practice

• Utilize Cohesion-6K for Arabic online discourse research.
• Apply the five discourse categories for nuanced content analysis.

Topics

• Cohesion-6K
• Arabic NLP
• Social Cohesion
• Online Conflict
• Facebook Discourse
• Computational Social Science

Best for: AI Scientist, NLP Engineer, Research Scientist

Related on AIssential

• Audience Engagement with Arabic Women's Social Empowerment and Wellbeing: A Decadal Corpus — The corpus provides a decadal view of Arabic women's social empowerment discourse on Facebook.
• Diálogos Tóxicos: Gatilhos e Padrões de Interação no Reddit Brasileiro — Online toxicity in Brazilian Portuguese Reddit discussions escalates from trigger comments, showing initial cohesion but eventual degradation.
• Discourse Relations and Connectives in Higher Text Structure — Local discourse analysis can reveal clues about higher text structure and global coherence.
• A Context-Aware Dataset for Stance Detection in Bioethical Controversies on Reddit — BioStance offers a large, context-aware dataset for stance detection in complex bioethical social media discourse.
• JobArabi: An Arabic Corpus and Analysis of Job Announcements from Social Media — JobArabi provides a unique Arabic social media corpus for analyzing labor market communication and sociolinguistic patterns.
• IMPACTeen: Intentions, Manipulation, Persuasion, Annotations, and Consequences in Teen Communication Dataset — The IMPACTeen dataset provides a multi-perspective, youth-contextualized resource for social influence research and language model development.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.