Mutual-enhanced Incongruity Learning Network for Multi-modal Sarcasm Detection
Published in AAAI 2023 (Oral), 2023
This paper proposes a joint model for multi-modal sarcasm detection, utilizing both local semantic-guided and global incongruity learning modules. It introduces a local semantic-guided incongruity learning module using IoU scores and ConceptNet5, and a global incongruity learning module based on attention mechanisms. The approach improves performance on sarcasm detection tasks by effectively addressing multimodal incongruity.
Recommended citation: Qiao, Yang, et al. "Mutual-enhanced incongruity learning network for multi-modal sarcasm detection." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 37. No. 8. 2023.
Download Paper