System Description Papers
Short papers (up to 4 pages excluding references and appendices) describing system approaches for the shared task, submitted via OpenReview and following EMNLP (ACL) templates.
NLP/CORE Unranked
ArabicNLP2026-DialectSentEval 2026
ArabicNLP 2026 Shared Task on Arabic Dialect Sentiment Analysis and Swapping
DialectSentEval 2026 is a shared task hosted within ArabicNLP 2026, focusing on sentiment analysis and sentiment swapping in Arabic dialects. It includes two subtasks: multi-class sentiment classification across five dialects and generative sentiment polarity inversion while preserving meaning, using curated datasets MDS-3 and MA’AKS Extended.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Shared Task Overview Paper
A separate paper summarizing the shared task, due after system papers and also following EMNLP formatting guidelines.
Research areas in scope
01
Shared Task Subtasks
Arabic Dialect Sentiment AnalysisArabic Sentiment Swap
Policies worth checking twice
- Papers must follow EMNLP (ACL) LaTeX or Word templates
- Maximum 4 pages for main content (excluding references and appendices)
- Papers are not anonymous (non-blind review)
- Papers must cite the shared task overview paper
- At least one author must register for ArabicNLP 2026 / EMNLP 2026
- No revisions or author name changes allowed after the review period
- Authors may post papers on arXiv at any time
- Ethical concerns must be discussed in the paper
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026