This repository represents the code as well as the datasets (e.g., input, annotations, etc.) used in the interpreting verbal irony research. Please check the input jsonlines file sarcasm_twitter_rte_separate.jsonlines.
The input file has the following members.
- serial number (snum) appended with ''-i'', where i denotes the ith rephrases
- sarcastic messages (sarcasm_message)
- intended (literal) rephrases from the Turkers (literal_message). Also appended with the "-i"
- the nature of incongruity (i.e., explicit or implicit) as annotated (see the paper).
If you use the dataset please cite the following paper,
Main Reference:
Interpreting Verbal Irony: Linguistic Strategies and the Connection to the{T}ype of Semantic Incongruity Debanjan Ghosh, Elena Musi, Kartikeya Upasani, Smaranda Muresan. Proceedings of the Society for Computation in Linguistics. 2020. pages 82--93.