@inproceedings{butas-etal-2026-tucnlp,
title = "{TUCNLP} at {S}em{E}val-2026 Task 11: Neuro-Symbolic Content Stripping for Debiased Syllogistic Reasoning",
author = "Butas, Rafael and
Lapusan, Alex and
Lemnaru, Camelia and
Potolea, Rodica",
editor = "Kochmar, Ekaterina and
Ghosh, Debanjan and
North, Kai and
Komachi, Mamoru",
booktitle = "Proceedings of the 20th {I}nternational {W}orkshop on {S}emantic {E}valuation (2026)",
month = jul,
year = "2026",
address = "San Diego, California, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.semeval-1.423/",
pages = "3408--3421",
ISBN = "979-8-89176-414-9",
abstract = "In this paper, we present the solution submitted by TUCNLP at SemEval-2026 Task{\textasciitilde}11: Disentangling Content and Formal Reasoning in Large Language Models. The task requires predicting the formal validity of categorical syllogisms while minimizing susceptibility to content-driven biases in English and 11 additional languages. We show that a modestly-sized model (Qwen3-8B) can achieve near-perfect logical reasoning on the English validity-only subtask, and large reductions in content effect on multilingual and premise-retrieval variants, when augmented with a multi-stage neuro-symbolic pipeline: LLM-based content stripping with iterative error correction converts natural language to abstract categorical forms, a classical symbolic parser validates against the twenty-four Aristotelian syllogistic forms, and asymmetric confidence thresholds mediate between symbolic and neural decisions. Across the four subtasks (ST1 to ST4), our system achieves accuracy ranging from 91.1{\textbackslash}{\%} to 100{\textbackslash}{\%} and bias-penalized ranking scores ({\$}{\textbackslash}mathcal{\{}M{\}}{\$}) from 31.8 to 100.0, with the main bottleneck being overconfident neural predictions that bypass symbolic verification."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="butas-etal-2026-tucnlp">
<titleInfo>
<title>TUCNLP at SemEval-2026 Task 11: Neuro-Symbolic Content Stripping for Debiased Syllogistic Reasoning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Rafael</namePart>
<namePart type="family">Butas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alex</namePart>
<namePart type="family">Lapusan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Camelia</namePart>
<namePart type="family">Lemnaru</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rodica</namePart>
<namePart type="family">Potolea</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 20th International Workshop on Semantic Evaluation (2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Kochmar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Debanjan</namePart>
<namePart type="family">Ghosh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai</namePart>
<namePart type="family">North</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mamoru</namePart>
<namePart type="family">Komachi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-414-9</identifier>
</relatedItem>
<abstract>In this paper, we present the solution submitted by TUCNLP at SemEval-2026 Task~11: Disentangling Content and Formal Reasoning in Large Language Models. The task requires predicting the formal validity of categorical syllogisms while minimizing susceptibility to content-driven biases in English and 11 additional languages. We show that a modestly-sized model (Qwen3-8B) can achieve near-perfect logical reasoning on the English validity-only subtask, and large reductions in content effect on multilingual and premise-retrieval variants, when augmented with a multi-stage neuro-symbolic pipeline: LLM-based content stripping with iterative error correction converts natural language to abstract categorical forms, a classical symbolic parser validates against the twenty-four Aristotelian syllogistic forms, and asymmetric confidence thresholds mediate between symbolic and neural decisions. Across the four subtasks (ST1 to ST4), our system achieves accuracy ranging from 91.1\textbackslash% to 100\textbackslash% and bias-penalized ranking scores ($\textbackslashmathcal{M}$) from 31.8 to 100.0, with the main bottleneck being overconfident neural predictions that bypass symbolic verification.</abstract>
<identifier type="citekey">butas-etal-2026-tucnlp</identifier>
<location>
<url>https://aclanthology.org/2026.semeval-1.423/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>3408</start>
<end>3421</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TUCNLP at SemEval-2026 Task 11: Neuro-Symbolic Content Stripping for Debiased Syllogistic Reasoning
%A Butas, Rafael
%A Lapusan, Alex
%A Lemnaru, Camelia
%A Potolea, Rodica
%Y Kochmar, Ekaterina
%Y Ghosh, Debanjan
%Y North, Kai
%Y Komachi, Mamoru
%S Proceedings of the 20th International Workshop on Semantic Evaluation (2026)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, USA
%@ 979-8-89176-414-9
%F butas-etal-2026-tucnlp
%X In this paper, we present the solution submitted by TUCNLP at SemEval-2026 Task~11: Disentangling Content and Formal Reasoning in Large Language Models. The task requires predicting the formal validity of categorical syllogisms while minimizing susceptibility to content-driven biases in English and 11 additional languages. We show that a modestly-sized model (Qwen3-8B) can achieve near-perfect logical reasoning on the English validity-only subtask, and large reductions in content effect on multilingual and premise-retrieval variants, when augmented with a multi-stage neuro-symbolic pipeline: LLM-based content stripping with iterative error correction converts natural language to abstract categorical forms, a classical symbolic parser validates against the twenty-four Aristotelian syllogistic forms, and asymmetric confidence thresholds mediate between symbolic and neural decisions. Across the four subtasks (ST1 to ST4), our system achieves accuracy ranging from 91.1\textbackslash% to 100\textbackslash% and bias-penalized ranking scores ($\textbackslashmathcal{M}$) from 31.8 to 100.0, with the main bottleneck being overconfident neural predictions that bypass symbolic verification.
%U https://aclanthology.org/2026.semeval-1.423/
%P 3408-3421Markdown (Informal)
[TUCNLP at SemEval-2026 Task 11: Neuro-Symbolic Content Stripping for Debiased Syllogistic Reasoning](https://aclanthology.org/2026.semeval-1.423/) (Butas et al., SemEval 2026)
ACL