@inproceedings{foroutan-etal-2026-retrieval,
title = "Retrieval-Augmented {LLM}s and Encoder Models for Multi-Label Climate Disinformation Narrative Classification",
author = "Foroutan, Neda and
Tsiakalou, Alexandra and
Schmitt, Vera",
editor = "Rehm, Georg and
Dietze, Stefan and
Dessi, Danilo and
Maynard, Diana and
Schimmler, Sonja",
booktitle = "Proceedings of Natural Scientific Language Processing ({NSLP}) @ {LREC} 2026",
month = may,
year = "2026",
address = "Palma, Mallorca (Spain)",
publisher = "ELRA Language Resources Association (ELRA)",
url = "https://aclanthology.org/2026.nslp-1.22/",
doi = "10.63317/2sxk32q2vrqz",
pages = "225--234",
abstract = "The detection of climate misinformation narratives remains challenging due to label imbalance, hierarchical taxonomies, and the multi-label nature of real-world claims. Developing models that can reliably assign fine-grained narrative categories is therefore essential for scalable analysis of climate disinformation. We present our approach to multi-label climate misinformation narrative classification for ClimateCheck@NSLP 2026 Task 2. The task requires assigning one or more narrative categories, defined by the hierarchical CARDS taxonomy, to climate-related claims. We investigate both encoder-based transformers and decoder-only large language models (LLMs), comparing fine-tuning BERT-based models with prompt-based and retrieval-augmented instruction tuning strategies with Qwen3 model. To address data scarcity and label imbalance, we explore targeted augmentation using external CARDS-based resources as well as semantic similarity filtering. Our experiments show that augmentation improves encoder-based models, with ModernBERT achieving competitive performance at low computational cost. However, the strongest results are obtained using retrieval-augmented instruction tuning with Qwen3, which narrows the candidate narrative space prior to prediction. This approach achieves a Macro-F1 score of 59.72{\%} on the official test set, securing second place on the leaderboard. These findings demonstrate the effectiveness of retrieval-guided LLM adaptation for structured multi-label narrative classification while highlighting the continued relevance of efficient encoder-based models."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="foroutan-etal-2026-retrieval">
<titleInfo>
<title>Retrieval-Augmented LLMs and Encoder Models for Multi-Label Climate Disinformation Narrative Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Neda</namePart>
<namePart type="family">Foroutan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alexandra</namePart>
<namePart type="family">Tsiakalou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vera</namePart>
<namePart type="family">Schmitt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of Natural Scientific Language Processing (NSLP) @ LREC 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Georg</namePart>
<namePart type="family">Rehm</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stefan</namePart>
<namePart type="family">Dietze</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Danilo</namePart>
<namePart type="family">Dessi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Diana</namePart>
<namePart type="family">Maynard</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sonja</namePart>
<namePart type="family">Schimmler</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA Language Resources Association (ELRA)</publisher>
<place>
<placeTerm type="text">Palma, Mallorca (Spain)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>The detection of climate misinformation narratives remains challenging due to label imbalance, hierarchical taxonomies, and the multi-label nature of real-world claims. Developing models that can reliably assign fine-grained narrative categories is therefore essential for scalable analysis of climate disinformation. We present our approach to multi-label climate misinformation narrative classification for ClimateCheck@NSLP 2026 Task 2. The task requires assigning one or more narrative categories, defined by the hierarchical CARDS taxonomy, to climate-related claims. We investigate both encoder-based transformers and decoder-only large language models (LLMs), comparing fine-tuning BERT-based models with prompt-based and retrieval-augmented instruction tuning strategies with Qwen3 model. To address data scarcity and label imbalance, we explore targeted augmentation using external CARDS-based resources as well as semantic similarity filtering. Our experiments show that augmentation improves encoder-based models, with ModernBERT achieving competitive performance at low computational cost. However, the strongest results are obtained using retrieval-augmented instruction tuning with Qwen3, which narrows the candidate narrative space prior to prediction. This approach achieves a Macro-F1 score of 59.72% on the official test set, securing second place on the leaderboard. These findings demonstrate the effectiveness of retrieval-guided LLM adaptation for structured multi-label narrative classification while highlighting the continued relevance of efficient encoder-based models.</abstract>
<identifier type="citekey">foroutan-etal-2026-retrieval</identifier>
<identifier type="doi">10.63317/2sxk32q2vrqz</identifier>
<location>
<url>https://aclanthology.org/2026.nslp-1.22/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>225</start>
<end>234</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Retrieval-Augmented LLMs and Encoder Models for Multi-Label Climate Disinformation Narrative Classification
%A Foroutan, Neda
%A Tsiakalou, Alexandra
%A Schmitt, Vera
%Y Rehm, Georg
%Y Dietze, Stefan
%Y Dessi, Danilo
%Y Maynard, Diana
%Y Schimmler, Sonja
%S Proceedings of Natural Scientific Language Processing (NSLP) @ LREC 2026
%D 2026
%8 May
%I ELRA Language Resources Association (ELRA)
%C Palma, Mallorca (Spain)
%F foroutan-etal-2026-retrieval
%X The detection of climate misinformation narratives remains challenging due to label imbalance, hierarchical taxonomies, and the multi-label nature of real-world claims. Developing models that can reliably assign fine-grained narrative categories is therefore essential for scalable analysis of climate disinformation. We present our approach to multi-label climate misinformation narrative classification for ClimateCheck@NSLP 2026 Task 2. The task requires assigning one or more narrative categories, defined by the hierarchical CARDS taxonomy, to climate-related claims. We investigate both encoder-based transformers and decoder-only large language models (LLMs), comparing fine-tuning BERT-based models with prompt-based and retrieval-augmented instruction tuning strategies with Qwen3 model. To address data scarcity and label imbalance, we explore targeted augmentation using external CARDS-based resources as well as semantic similarity filtering. Our experiments show that augmentation improves encoder-based models, with ModernBERT achieving competitive performance at low computational cost. However, the strongest results are obtained using retrieval-augmented instruction tuning with Qwen3, which narrows the candidate narrative space prior to prediction. This approach achieves a Macro-F1 score of 59.72% on the official test set, securing second place on the leaderboard. These findings demonstrate the effectiveness of retrieval-guided LLM adaptation for structured multi-label narrative classification while highlighting the continued relevance of efficient encoder-based models.
%R 10.63317/2sxk32q2vrqz
%U https://aclanthology.org/2026.nslp-1.22/
%U https://doi.org/10.63317/2sxk32q2vrqz
%P 225-234
Markdown (Informal)
[Retrieval-Augmented LLMs and Encoder Models for Multi-Label Climate Disinformation Narrative Classification](https://aclanthology.org/2026.nslp-1.22/) (Foroutan et al., NSLP 2026)
ACL