A Review of Deep Sarcasm Detection Research Based on Incongruity Theory
Main Article Content
Keywords
sarcasm detection, incongruity theory, multimodal learning, deep learning, natural language processing
Abstract
As one of the most classic types of counterfactual discourse, the core feature of satire is “subtext”, which involves detecting the cognitive conflict of this “hidden meaning within the words”. This paper uses incongruity as a clue to review the related work on deep sarcasm detection and provides a corresponding classification summary. Specifically, after briefly introducing the psychological definition of sarcasm, this paper analyzes the three research perspectives (semantic perspective, emotional perspective, and factual conflict) currently used for modeling sarcasm computation, and further presents the challenges faced by the current methods. Finally, it summarizes how DNN models such as Transformer and graph networks implicitly or explicitly measure the inconsistency between these modalities; it analyzes the existing problems of poor interpretability, insufficient modality alignment, and lack of common sense integration in current models, and proposes the future development trend of integrating multimodal large models with cognitive science research.
References
- [1] Kužina, V., Petric, A. M., Barišić, M., et al. (2023). CASSED: Context-based approach for structured sensitive data detection. Expert Systems with Applications, 223, Article 119924.
- [2] Fu, C. Y., Chen, Z. K., & Pan, L. (2023). A sarcasm detection method based on the fusion of textual semantics and social behavioral information. Chinese Journal of Network and Information Security, 9(4).
- [3] Wen, C., Jia, G., & Yang, J. (2023). Dip: Dual incongruity perceiving network for sarcasm detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 2540–2550).
- [4] Li, S. X., Hu, H. J., & Liu, M. F. (2023). Social media multimodal sentiment analysis based on semantic consistency. China Sciencepaper, 18(3).
- [5] Yue, T., Mao, R., Wang, H., et al. (2023). KnowleNet: Knowledge fusion network for multimodal sarcasm detection. Information Fusion, 100, Article 101921.
- [6] Sun, S. Q., Wei, G. Y., & Wu, S. (2025). Sentiment analysis method integrating hypergraph enhancement and dual contrastive learning. Journal of Computer Engineering & Applications, 61(22).
- [7] Vedhaviyassh, D. R., Sudhan, R., Saranya, G., et al. (2022). Comparative analysis of easyocr and tesseractocr for automatic license plate recognition using deep learning algorithm. In 2022 6th International Conference on Electronics, Communication and Aerospace Technology (pp. 966–971). IEEE.
- [8] Long, Y. C., Gou, Z. N., Chen, Y. X., et al. (2025). Multi-task generative reconstruction dialogue emotion recognition based on large language models. Application Research of Computers, 42(7).
- [9] Pfeifer, E., Wulf, H., Metz, K., et al. (2024). Correlations between meaning in life and nature connectedness: German-language validation of two topic-related measures and practical implications. Spiritual Care, 13(3), 242–256.
- [10] Tan, Y. Y., Chow, C. O., Kanesan, J., et al. (2023). Sentiment analysis and sarcasm detection using deep multi-task learning. Wireless Personal Communications, 129(3), 2213–2237.
- [11] Šandor, D., & Bagić Babac, M. (2024). Sarcasm detection in online comments using machine learning. Information Discovery and Delivery, 52(2), 213–226.
- [12] Luo, J., Li, Y., Li, X., et al. (2025). PKME-MLM: A novel multimodal large model for sarcasm detection. Computers, Materials, & Continua, 83(1), 877.
Article Sidebar
This work is licensed under a Creative Commons Attribution 4.0 International License.