Neuroscience Paper Ranking Methodology
How the current seed leaderboard works, what is provisional, and what the full Kurate-style scoring system is designed to become.
Current status
Sample tournament - experimental
The current public ranking is a curated seed corpus with sample pairwise fixtures layered onto matched papers. It is useful for product shape, source provenance, and early browsing, but it is not a validated production tournament ranking.
Current sample mode
The live leaderboard uses source-backed paper metadata, curated seed ratings, and a small fixture tournament for matched papers.
Target ingestion
The target pipeline will collect candidates from OpenAlex, PubMed, Crossref, and curated ScienceDex datasets, with field-level provenance.
Target assessment
Structured assessments will rate evidence strength, reuse value, graph fit, field signal, limitations, and source coverage.
Target tournament
More pairwise comparisons will narrow confidence intervals and convert the sample tournament into a production tournament when convergence targets are met.
Ranking Definition
ScienceDex is ranking reusable neuroscience impact, not raw citation count. A strong paper should contribute reusable evidence through data, protocols, atlases, tools, standards, or methods that other labs can build on.
Evidence
Seed rating axis. Pairwise weights and validation checks are planned before this becomes final.
Reuse
Seed rating axis. Pairwise weights and validation checks are planned before this becomes final.
Graph fit
Seed rating axis. Pairwise weights and validation checks are planned before this becomes final.
Field signal
Seed rating axis. Pairwise weights and validation checks are planned before this becomes final.
Known Limitations
- No full ingestion pipeline is live yet.
- No model-generated assessments are live yet.
- Only a small sample pairwise history is live.
- No expert validation benchmark is live yet.