A guard against ambiguous sentiment for multimodal aspect-level sentiment classification

Yanjing Wang

doi:10.1016/j.ipm.2025.104375

A guard against ambiguous sentiment for multimodal aspect-level sentiment classification

2025-09-17

0

PRE

AI

DOI：10.1016/j.ipm.2025.104375

Original

Original request for help

Save

Abstract

En 中文

• Introduces a novel method to selectively utilize image information based on sentiment ambiguity. • Proposes an ambiguity learning mechanism to assess sentiment uncertainty in text. • Exhibits superior performance compared to existing approaches on MALSC benchmarks. • Adapts the method to LLMs by designing an ambiguity-aware prompting strategy. • Conducts extensive experiments to analyze the method’s effectiveness.

A guard against ambiguous sentiment for multimodal aspect-level sentiment classification

Abstract

Journal

Organization