The SemEval-2016 Stance Dataset


A dataset of tweets manually annotated for stance towards given target, target of opinion (opinion towards), and sentiment (polarity). The full dataset with associated tweet id information is available here.

Details about this dataset and experiments on automatic classification for stance and sentiment are available in these papers:

Stance and Sentiment in Tweets. Saif M. Mohammad, Parinaz Sobhani, and Svetlana Kiritchenko. Special Section of the ACM Transactions on Internet Technology on Argumentation in Social Media, 2017, 17(3).
Paper (pdf)   BibTeX  

Detecting Stance in Tweets And Analyzing its Interaction with Sentiment. Parinaz Sobhani, Saif M. Mohammad, and Svetlana Kiritchenko. In Proceedings of the Joint Conference on Lexical and Computational Semantics (*Sem), August 2016, Berlin, Germany.
Paper (pdf)   BibTeX 

The stance labels for this dataset were used in a shared task competition: SemEval-2016 Task 6: Detecting Stance in Tweets. Further details about the data and the stance detection task can be found at the task website. The SemEval data is available for download here.

Semeval-2016 Task 6: Detecting Stance in Tweets. Saif M. Mohammad, Svetlana Kiritchenko, Parinaz Sobhani, Xiaodan Zhu, and Colin Cherry. In Proceedings of the International Workshop on Semantic Evaluation (SemEval ’16). June 2016. San Diego, California.
Paper (pdf)    BibTeX    Task Website

Automatic Stance Detection System: Many of the same features used in the NRC-Canada sentiment system were also used in a stance-detection system that outperformed submissions from all 19 teams that participated in SemEval-2016 Task 6 (Mohammad et al., 2017). Our stance system is not publicly available yet. However, the AffectiveTweets Package can be used to generate feature vectors from a large number of affect lexicons similar to those used by us for stance and sentiment.