Coling-UniA at SciVQA 2025: few-shot example retrieval and confidence-informed ensembling for Multimodal Large Language Models

This paper describes our system for the SciVQA 2025 Shared Task on Scientific Visual Question Answering. Our system employs an ensemble of two Multimodal Large Language Models and various few-shot example retrieval strategies. The model and few-shot setting are selected based on the figure and question type. We also select answers based on the models’ confidence levels. On the blind test data, our system ranks third out of seven with an average F1 score of 85.12 across ROUGE-1, ROUGE-L, and BERTS. Our code is publicly available.

Metadaten
Author:	Christian Jaumann, Annemarie Friedrich ORCiD GND, Rainer Lienhart ORCiD GND
URN:	urn:nbn:de:bvb:384-opus4-1264811
Frontdoor URL	https://opus.bibliothek.uni-augsburg.de/opus4/126481
ISBN:	979-8-89176-265-7OPAC
Parent Title (English):	Proceedings of the Fifth Workshop on Scholarly Document Processing (SDP 2025), 31 July 2025, Vienna, Austria
Publisher:	Association for Computational Linguistics (ACL)
Place of publication:	Stroudsburg, PA
Editor:	Tirthankar Ghosal, Philipp Mayr, Amanpreet Singh, Aakanksha Naik, Georg Rehm, Dayne Freitag, Dan Li, Sonja Schimmler, Anita De Waard
Type:	Conference Proceeding
Language:	English
Year of first Publication:	2025
Publishing Institution:	Universität Augsburg
Release Date:	2025/11/26
First Page:	230
Last Page:	239
DOI:	https://doi.org/10.18653/v1/2025.sdp-1.21
Institutes:	Fakultät für Angewandte Informatik
	Fakultät für Angewandte Informatik / Institut für Informatik
	Fakultät für Angewandte Informatik / Institut für Informatik / Lehrstuhl für Maschinelles Lernen und Maschinelles Sehen
	Fakultät für Angewandte Informatik / Institut für Informatik / Lehrstuhl für Computerlinguistik
Dewey Decimal Classification:	0 Informatik, Informationswissenschaft, allgemeine Werke / 00 Informatik, Wissen, Systeme / 004 Datenverarbeitung; Informatik
Licence (German):	CC-BY 4.0: Creative Commons: Namensnennung

Open Access