Open access
Aug 2026
Improving few-shot stance detection with MINE: meta-prompting and in-context N-shot examples
This work presents MINE (Meta-prompting and In-context N-shot Examples), a simple, light-weight, training-free framework that first creates a target-specific LLM stance agent through meta-prompting, and then augments it with a small number of semantically retrieved examples.
Uwaila Ekhator, Francesca Spezzano
· International Journal of Dat... · 0 citations