@inproceedings{hu-etal-2026-cross,
title = "Cross-Lingual Semantic Fusion: Towards Universal Sentence Representations via {LLM}s",
author = "Hu, Zhen and
Liu, Zhu and
Liu, Ying",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the {A}sia-Pacific Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.aacl-long.196/",
pages = "4359--4374",
ISBN = "979-8-89176-439-2",
abstract = "Extracting sentence representations from pretrained large language models (LLMs) is a training-free approach that leverages their inherent semantic understanding. Existing methods primarily rely on prompt engineering to distill core semantics, but the resulting representations often retain language-specific features that hinder semantic tasks. Moreover, these approaches fail to exploit the cross-lingual capabilities of LLMs, remaining confined to monolingual settings. To address these limitations, we propose CLSF, a novel method that actively introduces diverse language perturbations when encoding the same sentence and then fuses the perturbed embeddings, effectively reducing language-specific biases and producing a more universal representation. CLSF is plug-and-play and can be integrated with existing prompt-based methods. Experiments on Semantic Textual Similarity (STS), multilingual alignment, and various downstream tasks show that CLSF consistently improves performance across different LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hu-etal-2026-cross">
<titleInfo>
<title>Cross-Lingual Semantic Fusion: Towards Universal Sentence Representations via LLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhu</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ying</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-439-2</identifier>
</relatedItem>
<abstract>Extracting sentence representations from pretrained large language models (LLMs) is a training-free approach that leverages their inherent semantic understanding. Existing methods primarily rely on prompt engineering to distill core semantics, but the resulting representations often retain language-specific features that hinder semantic tasks. Moreover, these approaches fail to exploit the cross-lingual capabilities of LLMs, remaining confined to monolingual settings. To address these limitations, we propose CLSF, a novel method that actively introduces diverse language perturbations when encoding the same sentence and then fuses the perturbed embeddings, effectively reducing language-specific biases and producing a more universal representation. CLSF is plug-and-play and can be integrated with existing prompt-based methods. Experiments on Semantic Textual Similarity (STS), multilingual alignment, and various downstream tasks show that CLSF consistently improves performance across different LLMs.</abstract>
<identifier type="citekey">hu-etal-2026-cross</identifier>
<location>
<url>https://aclanthology.org/2026.aacl-long.196/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>4359</start>
<end>4374</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Cross-Lingual Semantic Fusion: Towards Universal Sentence Representations via LLMs
%A Hu, Zhen
%A Liu, Zhu
%A Liu, Ying
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-439-2
%F hu-etal-2026-cross
%X Extracting sentence representations from pretrained large language models (LLMs) is a training-free approach that leverages their inherent semantic understanding. Existing methods primarily rely on prompt engineering to distill core semantics, but the resulting representations often retain language-specific features that hinder semantic tasks. Moreover, these approaches fail to exploit the cross-lingual capabilities of LLMs, remaining confined to monolingual settings. To address these limitations, we propose CLSF, a novel method that actively introduces diverse language perturbations when encoding the same sentence and then fuses the perturbed embeddings, effectively reducing language-specific biases and producing a more universal representation. CLSF is plug-and-play and can be integrated with existing prompt-based methods. Experiments on Semantic Textual Similarity (STS), multilingual alignment, and various downstream tasks show that CLSF consistently improves performance across different LLMs.
%U https://aclanthology.org/2026.aacl-long.196/
%P 4359-4374
Markdown (Informal)
[Cross-Lingual Semantic Fusion: Towards Universal Sentence Representations via LLMs](https://aclanthology.org/2026.aacl-long.196/) (Hu et al., AACL-IJCNLP 2026)
ACL
- Zhen Hu, Zhu Liu, and Ying Liu. 2026. Cross-Lingual Semantic Fusion: Towards Universal Sentence Representations via LLMs. In Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4359–4374, Hengqin, China. Association for Computational Linguistics.