Benchmark RadarS2M-Bench
S2M-Bench evaluates the task of reconstructing mind maps from lecture slides, comprising 12,774 slide pages from 24 university courses with expert-annotated mind maps. The evaluation framework integrates ground-truth comparison, structure conformity analysis, and VLM-as-a-Judge scoring.
- Released
- 2026-08-01
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Existing benchmarks do not address automatic generation and evaluation of mind maps from educational slides, a task requiring balance of local and global knowledge. S2M-Bench provides a systematic evaluation framework to advance intelligent education tools by enabling comparison of models on this task.
Motivation
Generating mind maps from lecture slides can help learners efficiently assimilate fragmented knowledge, promising substantial benefits for intelligent education.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.