AiSciVision: A Framework for Specializing Large Multimodal Models in Scientific Image Classification

arXiv 2024 33 method

TLDR

AiSciVision specializes large multimodal models for scientific image classification using visual RAG and agentic tool use, evaluated on three real-world datasets.

Reasoning

The paper presents a clear framework with interpretability and real-world evaluation, but its scope is limited to image classification rather than full scientific discovery or automation of research workflows.

Read-first score

Read-first score 33, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 21.

Recency 8%

75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=analysis,dataset

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=dataset

Topical relevance 42%

17.5

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Field roles

Bridge

Rank sensitivity

Stability: volatile; rank range: 99.

Keyword Scores

AI for scientific research

autonomous research agent

research automation

AI scientist

automated research

scientific discovery agent

automated scientific discovery

automated experimentation

experiment design agent

literature review agent

survey generation

paper writing agent

Resources

Paper