12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

SciMKG: A Multimodal Knowledge Graph for Science Education with Text, Image, Video and Audio

Tong Lu, Zhichun Wang, Y. Zhou, Yiming Guan, Zhiyong Bai, Junsheng Du · Proceedings of the AAAI Conference on Artificial Intelligence · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

6/10
Relevance
0/4
Quality (LMQS)
D
Evidence
0
Citations
0.00
FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1609/aaai.v40i18.38574

Methodology & findings

Study design

Artifact construction with automated framework development.

Sample

> 1000

Primary method

Automatic evaluation metrics (F1 score for concept extraction) and human evaluation for robustness assessment of multimodal alignment method.

Main result

The study demonstrates that "our method improves concept extraction F1 score by 9% over state-of-the-art baselines" and that "both automatic and human evaluations confirm the robustness of our multimodal alignment method." The constructed SciMKG encompasses "1,356 knowledge points, 34,630 multimodal concepts, and 403,400 relational triples" for Chinese K12 science education.

Reports effect sizes.

Research paradigm

Positivist/Empiricist

Author conclusions

The authors conclude that "SciMKG and our construction toolkit will be publicly released to support further research and applications in AI-driven education," indicating their framework successfully addresses the challenge of constructing multimodal educational knowledge graphs from open resources while ensuring semantic consistency across modalities.

Open questions raised

  • The paper identifies that "constructing multimodal educational knowledge graphs (EKGs) from diverse open educational resources remains a challenge due to the reliance on costly manual annotations and the lack of multimodal integration."
  • The authors identify that constructing multimodal educational knowledge graphs from diverse open educational resources remains challenging due to reliance on costly manual annotations and lack of multimodal integration, which their work aims to address.
  • The paper identifies that constructing multimodal educational knowledge graphs from diverse open educational resources remains a challenge due to reliance on costly manual annotations and lack of multimodal integration. Future directions include further research and applications in AI-driven education using the released toolkit.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 61%

Explore related topics

Scholarly InfrastructureDesign Science

Related papers

← Browse all papers