ITLC at SemEval-2026 Task 11: Normalization and Deterministic Parsing for Formal Reasoning in LLMs

TL;DR

This paper presents a structural abstraction and deterministic parsing method to improve reasoning accuracy in multilingual large language models, reducing content biases and achieving top rankings in SemEval-2026 Task 11.

Contribution

It introduces a novel approach that transforms syllogisms into logical forms and applies deterministic parsing, outperforming fine-tuning methods in multilingual reasoning tasks.

Findings

Achieved top-5 rankings across all SemEval-2026 Task 11 subtasks.

Significantly reduced content effects in reasoning tasks.

Provided a competitive alternative to complex fine-tuning methods.

Abstract

Large language models suffer from content effects in reasoning tasks, particularly in multi-lingual contexts. We introduce a novel method that reduces these biases through explicit structural abstraction that transforms syllogisms into canonical logical representations and applies deterministic parsing to determine validity. Evaluated on the SemEval-2026 Task 11 multilingual benchmark, our approach achieves top-5 rankings across all subtasks while substantially reducing content effects and offering a competitive alternative to complex fine-tuning or activation-level interventions.