tgoop.com/opendatascience/2714
Last Update:
🚀 We're excited to announce #SemEval2026 Task 3: DimABSA!
This year, we're introducing a new shared task on Dimensional Sentiment and Stance Analysis, designed to push sentiment analysis beyond simple polarity to richer, more expressive representations.
🔹 Track A — DimABSA
Focuses on Dimensional Aspect-Based Sentiment Analysis, where systems predict continuous valence–arousal (VA) values for specific aspects. This gives a more nuanced picture of emotion than just "positive" or "negative.".
- Languages: English, Japanese, Russian, Tatar, Ukrainian, Chinese
🔹 Track B — DimStance
Explores stance detection as a "stance-as-aspect" problem, modeling stance in the same continuous VA space — bridging sentiment and stance Analysis.
- Languages: English, German, Hausa, Kinyarwanda, Swahili, Twi, Chinese
💡 Why DimABSA & DimStance?
Traditional sentiment analysis captures only coarse, categorical judgments (e.g., positive/negative), missing the emotional richness found in human affect.
DimABSA adopts continuous valence–arousal representations inspired by psychological models of emotion, distinguishing not only how positive or negative a sentiment is, but also how intense or calm it feels.
This finer granularity opens new directions for research and applications:
- Detecting high-arousal misinformation or emotionally charged posts
- Differentiating mental health signals (e.g., anxiety vs. depression)
- Modeling emotion dynamics in dialogue and personalized, empathetic systems
- Bridging sentiment and stance analysis across domains like politics or environmental protection
🗓 Key Dates
Evaluation Start: January 10, 2026
Evaluation End: January 31, 2026
System Description Paper Due: February 2026
Camera Ready Due: April 2026
The SemEval Workshop 2026 will be co-located with #ACL2026 in San Diego.
📄 All details, datasets, and participation info:
👉 https://github.com/DimABSA/DimABSA2026
We're organizing this task together with:
Liang-Chih Yu • Shamsuddeen H. Muhammad, PhD • Idris Abdulmumin • Jonas Becker • Lung-Hao Lee • Jin Wang • Jan Philip Wahle • Terry Ruas • Alexander Panchenko • Kai-Wei Chang • Saif M Mohammad
A huge thanks to this incredible team for their collaboration and ideas — it's been amazing shaping this together.
If you're working on sentiment analysis, stance detection, affective computing, or emotion modeling, we'd love to have you join us.
See you at SemEval 2026! 🌍💬
BY Data Science by ODS.ai 🦜
Share with your friend now:
tgoop.com/opendatascience/2714