iSydney v2.7 · SELF-OBSERVATION OF SELF-OBSERVATION

Gap Calibration Lab

It is not enough for iCIDM to generate candidate unknowns. The institution must later ask: Was I right to think this was a gap?

Candidate Gap → Human Review → Outcome → Reality Return → Compare Prediction with Outcome → Calibration → Better Future Gap Detection

Six adjudication outcomes

TRUE_SCIENTIFIC_UNKNOWNEvidence review confirms a meaningful unanswered scientific question.

CAPABILITY_GAPThe question is valid and the institution lacks or lost a relevant capability.

MEMORY_RETRIEVAL_GAPThe knowledge/capability existed, but the system failed to retrieve or connect it.

CLASSIFICATION_GAPThe apparent gap arose from labels, terminology, metadata or taxonomy.

APPROPRIATE_SPECIALISATIONThe sparse relationship is expected and scientifically appropriate.

NO_GAPReview shows the candidate was not a meaningful gap.

The calibration register starts open

All 9 current Question Passports begin as PENDING_REVIEW. No synthetic answer is filled in. Real review results become institutional learning data.

ID	Signal	Question	Outcome
ICIDM-UNASKED-001	RELATIONSHIP_SPARSE	Why is the Cryptococcus × Whole-genome sequencing relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-002	RELATIONSHIP_SPARSE	Why is the Arboviruses × Clinical trial / therapy relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-003	RELATIONSHIP_SPARSE	Why is the Aspergillus × Whole-genome sequencing relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-004	RELATIONSHIP_SPARSE	Why is the Staphylococcus aureus / MRSA × Serology / immunology relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-005	RELATIONSHIP_SPARSE	Why is the Bordetella pertussis × Clinical trial / therapy relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-006	RELATIONSHIP_SPARSE	Why is the Influenza / respiratory viruses × Whole-genome sequencing relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-007	RELATIONSHIP_SPARSE	Why is the Streptococcus pneumoniae × Bioinformatics / modelling relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-008	RELATIONSHIP_SPARSE	Why is the Influenza / respiratory viruses × Bioinformatics / modelling relationship sparse in the publication corpus, and is this a true capability gap, a terminology/classification gap, or simply an unrepresented relationship?	PENDING
ICIDM-UNASKED-009	STALE_OR_DORMANT	The Streptococcus pneumoniae × PCR / molecular diagnostics relationship was recurrent but has not appeared recently in the corpus. Is the capability obsolete, absorbed into newer methods, dormant, or simply relabelled?	PENDING

What the system should learn

After enough reviewed cases, iCIDM can measure separate rates such as true-unknown yield, memory-gap rate, classification-gap rate and false-gap rate. These should remain separate; a single “intelligence score” would hide why the system succeeds or fails.

Governance: calibration improves the question-generation mechanism; it must not suppress rare or unconventional questions merely because earlier candidates were often false positives.

A new institutional memory object

Lesson Passport: every adjudicated candidate should preserve what fooled the system, what evidence resolved it, whether metadata changed, whether an experiment was necessary, and which future rule should be improved.

UNKNOWN TAXONOMY · U1–U5

Gap calibration now has an action grammar

U1 calibrates representation; U2 preserves multiplicity; U3 removes false requirements; U4 seeks evidence; U5 can generate a Question Passport after review.

Use Unknown Gate →