PC-MNet: Dual-Level Congruity Modeling for Multimodal Sarcasm Detection via Polarity-Modulated Attention
Maoheng Li, Ling Zhou, Xiaohua Huang, Rubing Huang, Wenming Zheng, Guoying Zhao · 2026
Read PC-MNet: Dual-Level Congruity Modeling for Multimodal Sarcasm Detection via Polarity-Modulated Attention by Maoheng Li, Ling Zhou, Xiaohua Huang, Rubing Huang, Wenming Zheng, Guoying Zhao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-02447.txt.
Description
Maoheng Li and five co-authors build a system to detect sarcasm when words and nonverbal cues point in opposite ways. Sarcasm is hard for machines because the text may sound warm while a face or voice shows the reverse. Past systems often mixed the two streams late and in a uniform way, which can blur the clash that marks sarcasm.
Their model, PC-MNet, checks fit at two levels and lets a polarity signal steer attention. A routing step fuses only the most useful evidence, guided by a prior graph and by training that stresses contrast. On the MUStARD bench and on balanced versions built to cut false cues, the system beat the strongest prior multimodal baseline on a key score.
Catalog record
| CREATOR | Maoheng Li, Ling Zhou, Xiaohua Huang, Rubing Huang, Wenming Zheng, Guoying Zhao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.02447