AAA at SemEval-2026 Task 10: Psycholinguistic Conspiracy Marker Extraction and Detection

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Advanced, quick

Summary

AAA's study for SemEval-2026 Task 10 addresses psycholinguistic conspiracy marker extraction and detection within social media texts. The research focuses on token-level extraction and sentence-level conspiracy classification. It proposes a novel classification method that integrates semantic encoding with large language model reasoning and generation. Specifically, semantic features are extracted using DeBERTa-v3, and explanatory reasoning text is generated by ConspEmoLLM-v2. These two outputs are then combined to enhance the model's ability to recognize implicit conspiratorial logic. The study also includes a systematic comparison of several mainstream pre-trained models for the extraction subtask, providing baseline performance analysis.

Key takeaway

For NLP Engineers developing systems to detect subtle conspiratorial content on social media, this research suggests a powerful hybrid approach. You should consider integrating semantic encoding models like DeBERTa-v3 with large language models such as ConspEmoLLM-v2 for enhanced recognition of implicit conspiratorial logic. This method offers a robust strategy beyond traditional classification.

Key insights

Combining semantic encoding with LLM reasoning enhances detection of implicit conspiratorial logic in social media texts.

Principles

• Semantic encoding and LLM reasoning improve conspiracy detection.
• Token-level extraction aids sentence-level classification.

Method

Semantic features are extracted using DeBERTa-v3, explanatory reasoning text is generated via ConspEmoLLM-v2, then combined for classification.

In practice

• Apply DeBERTa-v3 for semantic feature extraction.
• Utilize ConspEmoLLM-v2 for reasoning text generation.
• Compare pre-trained models for token-level extraction.

Topics

• Conspiracy Detection
• Psycholinguistics
• Large Language Models
• Semantic Encoding
• DeBERTa-v3
• ConspEmoLLM-v2
• Social Media Analysis

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• mdok-style at SemEval-2026 Task 10: Finetuning LLMs for Conspiracy Detection — Data augmentation and self-training can effectively finetune LLMs for specialized text classification tasks with limited data.
• Exploring Knowledge Graphs for Automatic Fake News Detection in Portuguese — Knowledge Graphs offer an explainable method for automated fake news detection by modeling news as factual events.
• The Geometry of Truth: Detecting LLM Hallucinations with Geometric Algebra — Type II hallucinations are context contradictions, dangerous due to their plausibility and standard similarity metric blindness.
• An End-to-End Hybrid Framework for Rumour Detection in Low-Resources Algerian Dialect — A hybrid framework combining transformer embeddings with classical classifiers enables effective rumour detection in low-resource Algerian dialect.
• LLM-based Detection of Manipulative Political Narratives — LLM-driven pipelines can effectively identify and cluster manipulative political narratives in raw social media data.
• Sem-Detect: Semantic Level Detection of AI Generated Peer-Reviews — AI-generated peer reviews converge semantically, while human reviews, even LLM-refined, retain unique ideas.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.