A new, manually labeled dataset for the novel task of Sarcasm Target Identification.
There are two files - snippets.xlsx (224 book snippets) and tweets.xslx (506 tweets)
In both the files, each unit is a sarcastic text (snippet or tweet) annotated for the sarcasm target - the subset of words in the text which indicate the entity or the situation being ridiculed. In case the target is not found within the text, a fall-back label of 'Outside' is expected.
If using the snippets data, please cite: Joshi, Aditya, Pranav Goel, Pushpak Bhattacharyya, and Mark Carman. "Sarcasm target identification: Dataset and an introductory approach." In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018). 2018.
For tweets data: An extended version of that dataset is available at : http://www.alta.asn.au/events/sharedtask2019/description.html; Citation: Molla, Diego, & Joshi, Aditya (2019). Overview of the ALTA 2019 shared task: sarcasm target identification. In M. Mistica, M. Piccardi, & A. MacKinlay (Eds.), Proceedings of the 17th Annual Workshop of the Australasian Language Technology Association (pp. 192–196). Melbourne, VIC: Australasian Language Technology Association.