Emotion-stimulus data to accompany this paper:

Diman Ghazi, Diana Inkpen & Stan Szpakowicz (2015). “Detecting Emotion Stimuli in Emotion-Bearing Sentences”. Proceedings of the 16th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing 2015), Cairo, Egypt.

There are two files in the archive.

  1. The Emotion Cause dataset contains 820 sentences with both an emotion cause and an emotion tag. The tags are in the XML format: <cause> marks the beginning of the emotion stimulus span and <\cause> marks the end of the span.

    Example: <happy>These days he is quite happy <cause>travelling by trolley<\cause> . <\happy>

    The tags indicates the cause span and the emotion of the sentence — Happiness here.

  2. The No Cause dataset contains sentences marked with an emotion tag, which do not carry any emotion cause. This dataset contains 1594 instances marked with their emotion tag.

    Example: <anger>Bernice was so angry she could hardly speak . <\anger>

    This sentence belongs to the Anger emotion class, but the cause of the emotion is not mentioned in the sentence.

In both datasets, each sentence belongs to one of seven emotions: Happiness, Sadness, Anger, Fear, Surprise, Disgust and Shame.