@inproceedings{luo-etal-2026-beyond-correlation,
title = "Beyond Correlation-Driven Fusion: Sample-Adaptive Causal Evidence Learning for Multimodal Sentiment Analysis",
author = "Luo, Gaifang and
Zhang, Hao and
Liu, Yanan and
Tan, Haomin and
Wu, Guochuan and
Xu, Dan",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.2319/",
pages = "49799--49811",
ISBN = "979-8-89176-462-0",
abstract = "Existing multimodal sentiment analysis (MSA) methods rely on representation learning and cross-modal fusion. However, many follow a correlation-driven paradigm that may encode lexical shortcuts, visual biases, and non-causal correlations into predictions. To address this, we propose a Consistency-Conflict Fusion with Sample-Adaptive Causal Evidence (CCF-SACE) framework, revisiting MSA from a representation-level causal modeling perspective. First, encoders extract textual and visual features, and an adaptor projects them into a shared space. The framework then performs sample-adaptive cross-modal causal evidence selection, where textual causal evidence is selected under visual context and visual causal evidence under textual context. The remaining contexts are modeled as non-causal proxies, allowing the model to separate emotion-supportive evidence from shortcut-related information. CCF-SACE further introduces a covariance-based consistency-conflict-aware fusion module that constructs an evidence-level mediator by modeling cross-modal consistency and discrepancy rather than directly fusing all selected cues. Since the residual proxy stream may still carry label-correlated shortcut information, we introduce non-causal proxy suppression and counterfactual sensitivity objectives to weaken its standalone predictability and enlarge the discrepancy between evidence-mediated and proxy-only predictions. Experiments on five benchmarks show consistent improvements over state-of-the-art methods, especially on fine-grained affective recognition."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="luo-etal-2026-beyond-correlation">
<titleInfo>
<title>Beyond Correlation-Driven Fusion: Sample-Adaptive Causal Evidence Learning for Multimodal Sentiment Analysis</title>
</titleInfo>
<name type="personal">
<namePart type="given">Gaifang</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haomin</namePart>
<namePart type="family">Tan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guochuan</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dan</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Existing multimodal sentiment analysis (MSA) methods rely on representation learning and cross-modal fusion. However, many follow a correlation-driven paradigm that may encode lexical shortcuts, visual biases, and non-causal correlations into predictions. To address this, we propose a Consistency-Conflict Fusion with Sample-Adaptive Causal Evidence (CCF-SACE) framework, revisiting MSA from a representation-level causal modeling perspective. First, encoders extract textual and visual features, and an adaptor projects them into a shared space. The framework then performs sample-adaptive cross-modal causal evidence selection, where textual causal evidence is selected under visual context and visual causal evidence under textual context. The remaining contexts are modeled as non-causal proxies, allowing the model to separate emotion-supportive evidence from shortcut-related information. CCF-SACE further introduces a covariance-based consistency-conflict-aware fusion module that constructs an evidence-level mediator by modeling cross-modal consistency and discrepancy rather than directly fusing all selected cues. Since the residual proxy stream may still carry label-correlated shortcut information, we introduce non-causal proxy suppression and counterfactual sensitivity objectives to weaken its standalone predictability and enlarge the discrepancy between evidence-mediated and proxy-only predictions. Experiments on five benchmarks show consistent improvements over state-of-the-art methods, especially on fine-grained affective recognition.</abstract>
<identifier type="citekey">luo-etal-2026-beyond-correlation</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2319/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>49799</start>
<end>49811</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Correlation-Driven Fusion: Sample-Adaptive Causal Evidence Learning for Multimodal Sentiment Analysis
%A Luo, Gaifang
%A Zhang, Hao
%A Liu, Yanan
%A Tan, Haomin
%A Wu, Guochuan
%A Xu, Dan
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F luo-etal-2026-beyond-correlation
%X Existing multimodal sentiment analysis (MSA) methods rely on representation learning and cross-modal fusion. However, many follow a correlation-driven paradigm that may encode lexical shortcuts, visual biases, and non-causal correlations into predictions. To address this, we propose a Consistency-Conflict Fusion with Sample-Adaptive Causal Evidence (CCF-SACE) framework, revisiting MSA from a representation-level causal modeling perspective. First, encoders extract textual and visual features, and an adaptor projects them into a shared space. The framework then performs sample-adaptive cross-modal causal evidence selection, where textual causal evidence is selected under visual context and visual causal evidence under textual context. The remaining contexts are modeled as non-causal proxies, allowing the model to separate emotion-supportive evidence from shortcut-related information. CCF-SACE further introduces a covariance-based consistency-conflict-aware fusion module that constructs an evidence-level mediator by modeling cross-modal consistency and discrepancy rather than directly fusing all selected cues. Since the residual proxy stream may still carry label-correlated shortcut information, we introduce non-causal proxy suppression and counterfactual sensitivity objectives to weaken its standalone predictability and enlarge the discrepancy between evidence-mediated and proxy-only predictions. Experiments on five benchmarks show consistent improvements over state-of-the-art methods, especially on fine-grained affective recognition.
%U https://aclanthology.org/2026.findings-emnlp.2319/
%P 49799-49811
Markdown (Informal)
[Beyond Correlation-Driven Fusion: Sample-Adaptive Causal Evidence Learning for Multimodal Sentiment Analysis](https://aclanthology.org/2026.findings-emnlp.2319/) (Luo et al., Findings 2026)
ACL