A Deep Prompt-Based Chain-of-Thought Approach to Harmful Euphemism Detection in Social Networks
Siyu Xie, Gang Zhou, Haizhou Wang · 2026
Read A Deep Prompt-Based Chain-of-Thought Approach to Harmful Euphemism Detection in Social Networks by Siyu Xie, Gang Zhou, Haizhou Wang (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/pmc-13205860.txt.
Description
Siyu Xie, Gang Zhou, and Haizhou Wang hunt coded slurs on social networks in a 2026 paper. Harmful euphemisms hide slurs and bait in soft words, which stir bullying and group bias. Chinese cases face three blocks: few labeled sets, weak reasoning in small models, and slow large models.
The team built a large, fine-labeled set from platform posts. Their frame joins prompt-led step reasoning with contrast learning across heads. Outside knowledge from large models enriches the word senses the small model sees. A fusion stage merges many sense channels with shifting weights. The aim is fast, sharp spotting of veiled harm at platform scale.
Catalog record
| CREATOR | Siyu Xie, Gang Zhou, Haizhou Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 102 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |