Nice perfume. How long did you marinate in it? Multimodal Sarcasm Explanation

TL;DR

This paper introduces MuSE, a new dataset and model for generating natural language explanations of sarcasm in multimodal posts, combining images and text, to improve interpretability of sarcasm detection.

Contribution

It proposes the novel task of Multimodal Sarcasm Explanation (MuSE), creates the first dataset with explanations, and benchmarks a Transformer-based model for this purpose.

Findings

The model outperforms baselines across multiple metrics.

Human evaluation shows moderate agreement (Fleiss' Kappa 0.4).

Empirical results demonstrate the effectiveness of cross-modal attention.

Abstract

Sarcasm is a pervading linguistic phenomenon and highly challenging to explain due to its subjectivity, lack of context and deeply-felt opinion. In the multimodal setup, sarcasm is conveyed through the incongruity between the text and visual entities. Although recent approaches deal with sarcasm as a classification problem, it is unclear why an online post is identified as sarcastic. Without proper explanation, end users may not be able to perceive the underlying sense of irony. In this paper, we propose a novel problem -- Multimodal Sarcasm Explanation (MuSE) -- given a multimodal sarcastic post containing an image and a caption, we aim to generate a natural language explanation to reveal the intended sarcasm. To this end, we develop MORE, a new dataset with explanation of 3510 sarcastic multimodal posts. Each explanation is a natural language (English) sentence describing the hidden irony. We benchmark MORE by employing a multimodal Transformer-based architecture. It incorporates a cross-modal attention in the Transformer's encoder which attends to the distinguishing features between the two modalities. Subsequently, a BART-based auto-regressive decoder is used as the generator. Empirical results demonstrate convincing results over various baselines (adopted for MuSE) across five evaluation metrics. We also conduct human evaluation on predictions and obtain Fleiss' Kappa score of 0.4 as a fair agreement among 25 evaluators.