Arabic_Emotion_Datasets

• Paper Title: Enabling Deep Learning of Emotion With First-Person Seed Expressions

• Authors: Hassan Alhuzali, Muhammad Abdul-Mageed & Lyle Ungar.

• Email: halhuzali@alumni.ubc.ca & muhammad.mageeed@ubc.ca

Datasets description

Lama-dataset (Gold):

• Each line contains 3 comma separated fields (CSV), with the header: "tweet_id","emotion_relevance","emotion_category".
• "tweet_id": The Twitter id of the tweet
• "emotion_relevance:" Whether the tweet carries emotion or not (relevant to the task or not; binary "YES"/"NO")
• "emotion_category": The category of emotion the tweet carries.
• Note: Tweets that have "emotion_relevance"=NO are not considered for the "emotion_category" classification task. As such, the label under "emotion_category" is the suspected label if the "emotion_relevance" were to be equal to YES.
• Filename: "Lama_dataset.csv".

Lama-Dist (Distant Supervision):

• Each line contains 2 comma separated fields (CSV), with the header: "tweet_id","emotion_category".
• "tweet_id": The Twitter id of the tweet
• "emotion_category": The category of emotion the tweet carries.
• Filename: "Lama_dist_dataset.csv".

Replication

• We provide our data splits for Lama-dataset (Gold) as described in the paper.
• Link: Dataset-split

License & co

• The dataset is released exclusively for research purposes. For other uses, permission must be acquired from the authors.
• Please cite the paper if you use our data.

@article{Alhuzali2018,
  title = {{Enabling Deep Learning of Emotion With First-Person Seed Expressions}},
  author = {Alhuzali, Hassan and Abdul-Mageed, Muhammad and Ungar, Lyle},
  pages = {25--35},
  url = {http://aclweb.org/anthology/W18-1104},
  year = {2018}
}