UFG-Semantic at SemEval-2026 Task 6: CLARITY - Unmasking Political Question Evasions

AbstractWe propose an approach for Task 6: CLARITY - Unmasking Political Question Evasions. We make use of data augmentation, supervised fine-tuning, and model benchmarking to detect and classify response ambiguity in political discourse. Building on well-founded theory on equivocation and leveraging recent advancements in language modeling, our system was structured based on question/answer (QA) pairs extracted from presidential interviews, and it was evaluated in Clarity-level Classification and Evasion-level Classification.