Two Stage Context Learning with Large Language Models for Multimodal Stance Detection on Climate Change

Lata Pangtey, Omkar Kabde, Shahid Shafi Dar, Nagendra Kumar · 2025

Read Two Stage Context Learning with Large Language Models for Multimodal Stance Detection on Climate Change by Lata Pangtey, Omkar Kabde, Shahid Shafi Dar, Nagendra Kumar (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-08024.txt.

Description

Social media posts mix words with images, so judging a post's stance on climate change needs both. Lata Pangtey and 3 co-authors present a two-stage multimodal framework in a 2025 research paper. First, a large language model pulls stance-relevant summaries from the source text. In parallel, a domain-aware caption generator describes the image in topic context.

A specialized transformer then models the source text, the image reading, and the reply text together. The team evaluates on the MultiClimate benchmark, which pairs video frames with transcripts. Their method beat existing state-of-the-art approaches on accuracy and related scores. The design shows how hierarchical fusion, combining each mode at the right level, steadies stance detection. Such tools help researchers track public opinion on contested issues at scale.

Catalog record

CREATORLata Pangtey, Omkar Kabde, Shahid Shafi Dar, Nagendra Kumar
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-08024

Browse the full library · About the Tower of Babel