@inproceedings{goel-etal-2025-target,
    title = "Target-Augmented Shared Fusion-based Multimodal Sarcasm Explanation Generation",
    author = "Goel, Palaash  and
      Chauhan, Dushyant Singh  and
      Akhtar, Md Shad",
    editor = "Chiruzzo, Luis  and
      Ritter, Alan  and
      Wang, Lu",
    booktitle = "Findings of the Association for Computational Linguistics: NAACL 2025",
    month = apr,
    year = "2025",
    address = "Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/Ingest-2025-COMPUTEL/2025.findings-naacl.472/",
    pages = "8480--8493",
    ISBN = "979-8-89176-195-7",
    abstract = "Sarcasm is a linguistic phenomenon that intends to ridicule a target (e.g., entity, event, or person) in an inherent way. Multimodal Sarcasm Explanation (MuSE) aims at revealing the intended irony in a sarcastic post using a natural language explanation. Though important, existing systems overlooked the significance of the target of sarcasm in generating explanations. In this paper, we propose a \textbf{T}arget-a\textbf{U}gmented sha\textbf{R}ed fusion-\textbf{B}ased sarcasm explanati\textbf{O}n model, aka. . We design a novel shared-fusion mechanism to leverage the inter-modality relationships between an image and its caption. assumes the target of the sarcasm and guides the multimodal shared fusion mechanism in learning intricacies of the intended irony for explanations. We evaluate our proposed model on the dataset. Comparison against multiple baselines and state-of-the-art models signifies the performance improvement of by an average margin of $+3.3\%$. Moreover, we explore LLMs in zero and one-shot settings for our task and observe that LLM-generated explanation, though remarkable, often fails to capture the critical nuances of the sarcasm. Furthermore, we supplement our study with extensive human evaluation on {\textquoteleft}s generated explanations and find them out to be comparatively better than other systems."
}