Statistics
The state of the benchmark
A live cross-section of the University of Scientific Workflow corpus — domains, difficulty, tools, environments, and workflow shape. Read it to find the gaps: the thin domains and missing difficulty tiers are exactly where your lab's next contribution counts most.
Tasks per domain
Where the corpus is deep — and where it is thin. Faded bars are domains with zero tasks.
Tasks by type
The proposal's coverage taxonomy — outcome known & measurable, open but still verifiable, or judged by a scientist's rubric.
Both axes verifiable — the outcome is quantitatively checkable and every step is scored against the scientist's ground-truth targets.
An open problem. The outcome stays quantitatively verifiable, but the workflow is not — an expected direction may be supplied, while execution strategy and per-step targets remain open.
An open problem whose outcome cannot be quantitatively verified, so it is judged by a scientist's rubric; the workflow is not verifiable either.
Difficulty distribution
Balance across Easy → Expert tiers.
- Easy00%
- Medium735%
- Hard840%
- Expert525%
Review status
Verified vs. in-flight contributions.
- Verified15%
- Under review315%
- Seed / sample1680%
Corpus growth
Cumulative task total, with newly registered tasks per month.
Most-used simulations
How many tasks reference each tool or solver.
Execution environments
Where the referenced tools run.
- HPC1743%
- Terminal1538%
- VM (GUI)38%
- Web API513%
Coverage shape
Task count per domain, as a coverage footprint.
Average workflow length
Mean number of atomic steps per task, by domain.
Domain × difficulty coverage
Every empty cell is an uncollected slice — pick one for your next submission.
Spotted a gap?
Several difficulty tiers within covered domains still sit empty. Bring the workflow behind your Nature-family paper to the benchmark — after expert review it becomes a task here.