AILS-NTUA at SemEval-2026 Task 10: Agentic LLMs for Psycholinguistic Marker Extraction and Conspiracy Endorsement Detection

TL;DR

A novel agentic LLM pipeline for SemEval-2026 Task 10 that jointly extracts psycholinguistic conspiracy markers and detects conspiracy endorsement and proposes Dynamic Discriminative Chain-of-Thought with deterministic anchoring to resolve semantic ambiguity and character-level brittleness.

Abstract

This paper presents a novel agentic LLM pipeline for SemEval-2026 Task 10 that jointly extracts psycholinguistic conspiracy markers and detects conspiracy endorsement. Unlike traditional classifiers that conflate semantic reasoning with structural localization, our decoupled design isolates these challenges. For marker extraction, we propose Dynamic Discriminative Chain-of-Thought (DD-CoT) with deterministic anchoring to resolve semantic ambiguity and character-level brittleness. For conspiracy detection, an "Anti-Echo Chamber" architecture, consisting of an adversarial Parallel Council adjudicated by a Calibrated Judge, overcomes the "Reporter Trap," where models falsely penalize objective reporting. Achieving 0.24 Macro F1 (+100\% over baseline) on S1 and 0.79 Macro F1 (+49\%) on S2, with the S1 system ranking 3rd on the development leaderboard, our approach establishes a versatile paradigm for interpretable, psycholinguistically-grounded NLP.

Figures (15)

• Figure 1: System architecture. S1: DD-CoT Self-Refine extracts markers; a deterministic verifier anchors them to character offsets. S2: Anti-Echo Chamber with contrastive retrieval, forensic profiling, a Parallel Council, and a calibrated Judge.
• Figure 2: Contrastive few-shot retrieval architecture.
• Figure 3: GEPA prompt optimization workflow.
• Figure 4: Juror Ablation Study (LOO). Removing any single persona significantly degrades performance, validating the necessity of the full four-juror council.
• Figure 5: Number of marker types in the dataset.