Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Christos Tzouvaras, Konstantinos Skianis, Athanasios Voulodimos

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.

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

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.
Paper Structure (28 sections, 6 equations, 5 figures, 9 tables)

This paper contains 28 sections, 6 equations, 5 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Task Setup
  4. Related Work
  5. System Overview
  6. Dual-Model Ensemble
  7. Deliberative Complexity Gating (DCG)
  8. Experimental Setup
  9. Implementation and Dependencies
  10. Inference Hyperparameters
  11. Prompting Strategy
  12. Evaluation Measures
  13. Results
  14. Evaluation
  15. Ablation Study
  16. ...and 13 more sections

Figures (5)

  • Figure 1: Task label taxonomy by thomas2024isaidthatdataset (evasion labels mapped to 3 clarity classes).
  • Figure 2: Architecture of our two-stage pipeline. Stage 1 runs Grok and Gemini with $k=5$ self-consistency (SC), maps evasion to clarity, and applies asymmetric weighted voting. Stage 2 (DCG) uses Gemini response length and Grok consistency to gate uncertain cases.
  • Figure A1: DCG sensitivity to percentile choice. Macro-F1 as $\theta_1$ moves from the 5th to the 95th percentile on Development and Evaluation.
  • Figure A2: Gemini response length by gold clarity class. Ambivalent samples are generally longer than clear classes on both splits.
  • Figure A3: Agreement $\times$ consistency heatmap (post-DCG). Accuracy by Grok--Gemini agreement and Grok self-consistency bins on Development and Evaluation sets. The dashed red outline highlights disagreement with medium/low-consistency cells as a diagnostic focus region.