Stance detection aims to detect whether a given text is favorable or unfavorable to a given target entity.
https://github.com/sheffieldnlp/stance-conditional/
This corpus has been proposed for the SemEval 2016 Stance Detection for Twitter shared task. It consists of two subtasks, Task A and Task B. In Task A the goal is to detect the stance of tweets towards targets given labelled training data for five test targets (Climate Change is a Real Concern, Feminist Movement, Atheism, Legalization of Abortion and Hillary Clinton). In Task B, the goal is to detect stance with respect to an unseen target (Donald Trump) for which labeled data are not provided.