A paper appears in Featured research when a manual assessment meets the defined threshold. The assessment covers problem importance, scientific novelty, breadth and reusability, evidence strength, and personal contribution. Venue format is reported factually and is not scored. Newly synchronized records remain unranked until review.
View assessment criteria
- Problem importance
- 25%
- Scientific novelty
- 25%
- Breadth and reusability
- 20%
- Evidence strength
- 20%
- Personal contribution
- 10%
Minimum score: 4.2/5 for Featured research. Observed impact, including independent use, awards, successful replications, and field-normalized citation impact, is assessed later and kept separate from the assessment made at publication time. Article-level scores remain internal to avoid false precision. This is an editorial ranking for this personal site, not a universal claim about scientific value.
Machine-readable archive access
Library harvesters can retrieve every maintained record through the read-onlyOAI-PMH 2.0 repository. Agents can use thepublication index, and API clients can inspect theOpenAPI description. These interfaces are generated from the same validated records as the human-readable pages.
Publication result summaries. Each curated paper also exposes a small CSV, JSON, and Markdown summary of its reported table values. These files are for discovery, citation, and verification; they are not the underlying training or evaluation datasets and do not imply a public code release. The two external mirrors are grouped in one Hugging Face collection.