$R^3$: Reverse, Retrieve, and Rank for Sarcasm Generation with Commonsense Knowledge
This addresses the problem of generating high-quality sarcasm for natural language processing applications, representing an incremental improvement over prior methods.
The paper tackles sarcasm generation from non-sarcastic sentences by combining valence reversal and semantic incongruity using commonsense knowledge, resulting in a system that outperforms human annotators 34% of the time and a baseline 90% of the time in human evaluations.
We propose an unsupervised approach for sarcasm generation based on a non-sarcastic input sentence. Our method employs a retrieve-and-edit framework to instantiate two major characteristics of sarcasm: reversal of valence and semantic incongruity with the context which could include shared commonsense or world knowledge between the speaker and the listener. While prior works on sarcasm generation predominantly focus on context incongruity, we show that combining valence reversal and semantic incongruity based on the commonsense knowledge generates sarcasm of higher quality. Human evaluation shows that our system generates sarcasm better than human annotators 34% of the time, and better than a reinforced hybrid baseline 90% of the time.