Mu2STS: A Multitask Multimodal Sarcasm-Humor-Differential Teacher-Student Model for Sarcastic Meme Detection
- Publisher:
- SPRINGER INTERNATIONAL PUBLISHING AG
- Publication Type:
- Chapter
- Citation:
- Advances in Information Retrieval, 2024, 14610 LNCS, pp. 19-37
- Issue Date:
- 2024-01-01
Recently Added
Filename | Description | Size | |||
---|---|---|---|---|---|
978-3-031-56063-7_2.pdf | Published version | 1.65 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is new to OPUS and is not currently available.
Memes, a prevalent form of online communication, often express opinions, emotions, and creativity concisely and entertainingly. Amidst the diverse landscape of memes, the realm of sarcastic memes holds a unique position with its foundation in irony, mockery, satire, and messages that diverge from literal meanings. Detecting sarcasm in memes is challenging due to the intricate interplay between sarcasm and humor. While prior research has primarily concentrated on leveraging the relationship between sarcasm and humor for identifying sarcastic memes, our goal in this paper extends beyond establishing a fundamental connection between the two; instead, we aspire to unravel their distinct characteristics and nuances that differentiate sarcasm from humor. To accomplish this, we introduce a novel deep learning model, i.e., Mu2STS (MultitaskMultimodalSarcasm-Humor-DifferentialTeacher-Student), for sarcasm detection in memes, with a special focus on humor. To bolster Mu2STS, we have developed the SHMH (WARNING: This paper contains meme samples that are offensive in nature.) (Sarcasm-with-Humorous-Meme-in-Hindi) dataset, designed for detecting sarcasm and humor in memes written in the Hindi language, which is the first of its kind to the best of our knowledge. Our empirical evaluation, which includes both qualitative and quantitative analyses conducted on the SHMH dataset and some benchmark meme datasets, clearly illustrates the effectiveness of Mu2STS, which outperformed major state-of-the-art models. (The dataset and codes are available at https://www.iitp.ac.in/~ai-nlp-ml/resources.html.)
Please use this identifier to cite or link to this item: