← CIDM Research View · Check My CIDM →
See the work. Treat the people fairly. Keep the history useful.
This layer exists because a technically complete corpus can still represent people unfairly. CIDM colleagues should be able to find their real work, challenge wrong attribution or missing context, and see how past research connects to future questions.
1 · Visibility
Older work must not disappear simply because old indexing is weaker. Recent digital richness must not dominate the historical record.
2 · Proportion
Authorship, collaboration, laboratory support and institutional contribution are represented according to evidence. Publication count is not a staff ranking.
3 · Context
A paper is not just a citation. Where possible, the system preserves the scientific question, period, collaborators, evidence basis and uncertainty.
4 · Future value
Historical review should expose durable capabilities, unresolved questions and research lineages that new CIDM colleagues can continue.
Reverse audit — how we try to find our own mistakes
Forward review asks what a paper says. Reverse audit asks where our representation could be wrong. Public-facing claims are challenged first.
- Identity challenge: is this really the intended CIDM person/paper, or a same-name collision?
- Inclusion challenge: what exact affiliation, historical, or verified core-person evidence put the paper in the corpus?
- Evidence-depth challenge: does FULL / ABSTRACT / METADATA match what was actually inspected?
- Semantic challenge: did the summary overstate causality, novelty, priority, effect or contribution?
- Historical-fairness challenge: did data availability make one era/person look larger or smaller than reality?
- Future-value challenge: can this record help explain a research lineage, capability, unresolved question or next step?
Public-language rule
Priority words such as first, pioneered, led, definitive, established require unusually strong evidence. When evidence is weaker, prefer bounded language such as contributed to, within the reviewed corpus, or the available record indicates.
Current control boundary
Corpus: 1,947 exact publication objects. First-pass Deep iReview: 1,947/1,947 closed. Active depth: 258 FULL · 1,544 ABSTRACT · 145 METADATA. Human Gate: OPEN. Formal Accept: 0.
AI review is preliminary scientific assistance. Citation ≠ evidence; access ≠ review; review ≠ acceptance; author confirmation is representation evidence, not a truth verdict.
Name variants are preserved
Find-all without merge-all: My CIDM uses an identity-aware alias registry so one colleague can retrieve publications recorded under different name forms, initials or established variants. Alias expansion is a retrieval aid only; it never changes the original author string and never overrides paper-level identity/evidence review.