CSE-UOI at SemEval-2026 Task 6: A Two-Stage Heterogeneous Ensemble with Deliberative Complexity Gating for Political Evasion Detection

2026-03-12

AI Summary

This study addresses the automatic classification of response clarity in political interviews, categorizing utterances into three types: clear answers, ambiguous responses, or explicit evasions. To tackle this challenge, we propose a heterogeneous large language model ensemble framework that integrates self-consistency reasoning with a weighted voting mechanism. A novel post-processing module, Deliberative Complexity Gating (DCG), is introduced to dynamically refine borderline classifications by leveraging response length as a proxy for ambiguity. We also explore multi-agent debate as an alternative strategy to enhance reasoning. Evaluated on SemEval-2026 Task 6, our approach achieves a Macro-F1 score of 0.85, ranking third overall and demonstrating significantly improved detection of ambiguous responses.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Argumentation

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

This paper describes our system for SemEval-2026 Task 6, which classifies clarity of responses in political interviews into three categories: Clear Reply, Ambivalent, and Clear Non-Reply. We propose a heterogeneous dual large language model (LLM) ensemble via self-consistency (SC) and weighted voting, and a novel post-hoc correction mechanism, Deliberative Complexity Gating (DCG). This mechanism uses cross-model behavioral signals and exploits the finding that an LLM response-length proxy correlates strongly with sample ambiguity. To further examine mechanisms for improving ambiguity detection, we evaluated multi-agent debate as an alternative strategy for increasing deliberative capacity. Unlike DCG, which adaptively gates reasoning using cross-model behavioral signals, debate increases agent count without increasing model diversity. Our solution achieved a Macro-F1 score of 0.85 on the evaluation set, securing 3rd place.

Problem

• political evasion detection
• response clarity classification
• ambiguity detection
• political interviews
• non-reply identification

Innovation

• Heterogeneous Ensemble
• Deliberative Complexity Gating
• Self-Consistency
• Ambiguity Detection
• Multi-Agent Debate