Conflicts And Recovery¶

Scientific curation is incomplete when disagreement or missing evidence is hidden behind one convenient value. Bijux Pollenomics preserves competing claims, identifies their owners and scopes, and records the evidence required to resolve or safely bound the disagreement.

The domain language distinguishes conflict, qualification, recovery condition, posture, scope, and fact ownership.

Distinguish The Gap¶

State	Meaning	Appropriate response
absent upstream	the identified source does not provide the needed fact	retain the absence and seek another authoritative source if justified
not captured	the source may contain the fact, but the governed capture does not	recover the named artifact or field and record retrieval identity
captured but unparsed	useful source text exists but cannot be normalized safely	preserve text, improve a bounded parser only when semantics are known
ambiguous	several identities or interpretations remain plausible	keep alternatives and require a discriminating locator or review decision
conflicting	captured sources make incompatible claims	preserve both, compare scope and authority, and record the governing decision
unsupported for use	the fact may be real, but it does not satisfy the proposed product claim	qualify, use as context, defer, or refuse without weakening the gate
outside scope	evidence is valid but not a member of this geography or product	retain source evidence and account for the non-membership

Missing, false, excluded, and out of scope are not synonyms. Recovery begins only after the state is correctly classified.

Conflict Record¶

A conflict is reviewable when it preserves:

the governed subject and claim dimension;
every competing value at its original scope;
source family, release or accession, artifact, locator, and source wording;
whether each claim is sample-, site-, project-, paper-, or product-owned;
precision, dating basis, coordinate basis, and other material qualifiers;
the selected governing claim, if one is justified, and why;
the public effect and the evidence that would reopen the decision.

flowchart LR
    Left["claim A + locator + scope"] --> Compare{"authority and meaning review"}
    Right["claim B + locator + scope"] --> Compare
    Compare -->|one governs| Decision["select for declared use; retain both"]
    Compare -->|both compatible at different scopes| Qualify["preserve scoped claims"]
    Compare -->|unresolved| Queue["named recovery item"]
    Decision --> Descendants["re-evaluate affected products"]
    Qualify --> Descendants

A sample-owned date can govern that sample while a broader project interval remains valid context. The two claims need not be averaged or reduced to one unqualified field.

Classify Disagreement Before Resolving It¶

Not every unequal value is a contradiction. First identify what differs:

Disagreement class	Example	Correct treatment
identity conflict	one label maps plausibly to two samples	retain both candidates until a discriminating locator resolves identity
value conflict	two sample-owned coordinates claim different exact places	compare source authority, method, and precision; preserve the losing claim
scope difference	a project interval and a narrower sample interval differ	retain both at their natural scopes; use the sample claim for the sample
method difference	calibrated and uncalibrated BP values appear numeric but use different bases	refuse direct comparison or normalize only under a declared conversion contract
precision difference	region-level locality and exact supplied coordinate describe the same sample	retain both; do not treat the broader statement as contradictory
revision difference	a later source release changes a value	compare releases and lineage; do not overwrite the prior snapshot silently
role difference	one record is direct evidence and another is environmental context	preserve both roles; do not ask majority agreement to select an owner

flowchart TD
    Values["unequal claims"] --> Subject{"same typed subject?"}
    Subject -->|no| Identity["identity or join problem"]
    Subject -->|yes| Scope{"same dimension, scope, method, and revision?"}
    Scope -->|no| Plural["compatible scoped claims"]
    Scope -->|yes| Conflict["true claim conflict"]
    Identity --> Recovery["recover discriminating evidence"]
    Plural --> Preserve["preserve each natural scope"]
    Conflict --> Authority["resolve by fact ownership and evidence"]

Only the final branch is a value conflict. Resolving scope or identity first prevents curation from manufacturing disagreement through an invalid join.

Resolve By Authority And Scope¶

Conflict resolution follows evidence ownership before convenience:

confirm that the claims concern the same governed subject and dimension;
compare observation level, source locator, method, and precision;
identify the declared fact owner and the scope in which each source speaks;
determine whether the claims conflict or are compatible at different scopes;
select a governing value only when the evidence contract justifies it;
preserve every competitor and record the effect on dependent products.

Recency alone is not a resolution rule. Neither are majority vote, non-null preference, greater numeric precision, or agreement with the current map. Those shortcuts can select a value, but they cannot establish that it belongs to the correct object or claim.

Conflict Resolution Receipt¶

A resolved conflict remains auditable only when the resolution retains the comparison that justified it:

Receipt member	Required evidence
subject and dimension	one typed governed object and the exact fact under dispute
candidate claims	every value, source wording, locator, scope, method, precision, and revision
comparability finding	whether the candidates truly conflict or differ by identity, scope, role, method, or revision
authority finding	declared fact owner and why it applies to this subject and use
outcome	governing claim, scoped coexistence, qualification, or unresolved posture
retained competitor	losing or non-governing claims remain queryable rather than disappearing
descendant effect	decisions, memberships, counts, maps, and language re-evaluated from the outcome

If the retained record shows only the chosen value, the conflict has been hidden rather than resolved. If it shows both values without scope and owner, the disagreement has been stored but not interpreted.

Recovery Is Evidence Acquisition¶

Recovery items name a scientific deficit, not merely an unfinished file:

Recovery target	Completion evidence
supporting material	content-identified artifact, paper link, manifest membership, and retrieval context
sample identity	source-owned label, stable repository identity, project relation, and ambiguity disposition
locality	sample-owned wording, source locator, site relation, hierarchy, and resolution posture
chronology	source wording, owner, normalized point or interval when defensible, basis, and precision
coordinate	source pair or documented named-place resolution, method, basis, confidence, and mapping posture
product admission	rerun decision with passed rules, updated accountability, and descendant review

Finding a plausible value on the web is not completion. The value becomes governed evidence only when its identity, source, locator, ownership, meaning, and downstream impact are recorded.

A Recovery Attempt Does Not Change The Gap By Itself¶

Acquisition outcome and scientific recovery posture are separate states:

Attempt outcome	Scientific interpretation
service unavailable or authentication refused	recovery blocked; the governed gap remains unchanged
response succeeds but contains no identified target member	searched scope yielded no recoverable member; absence is bounded to that request and revision
artifact recovered but target relation is ambiguous	material acquired; claim remains unresolved pending identity evidence
value recovered without a stable locator	candidate information only; not governed evidence
named artifact and discriminating relation recovered	reevaluate the affected claim and descendants; do not auto-admit

Repeated failure cannot turn a gap into source absence, and successful transport cannot turn a plausible value into a resolved claim. The recovery record retains the request identity and outcome separately from the decision it was intended to change.

Recovery Priority¶

Recovery order is driven by scientific impact and resolvability, not by which missing field is easiest to fill. A useful queue records:

Dimension	Question
affected authority	does the gap sit in a governing record or only a derived view?
descendant reach	which comparisons, maps, counts, and reports can change?
claim severity	does the gap risk false identity, false precision, or only reduced context?
recoverability	is a named archive object, supplement, field, or relation available?
decision stability	could recovery change admission, or only strengthen provenance?
verification cost	which semantic and membership comparisons are required afterward?

This makes high-impact identity and relation failures visible even when a lower-impact metadata field would be faster to complete.

Recovery Loop¶

stateDiagram-v2
    [*] --> Classified: gap or conflict identified
    Classified --> Acquired: supporting material captured
    Acquired --> Curated: identity and claims reviewed
    Curated --> Rejected: evidence remains insufficient
    Rejected --> Classified: retain reason and next condition
    Curated --> Accepted: governing decision supported
    Accepted --> Rebuilt: affected descendants regenerated
    Rebuilt --> Verified: semantic and membership diffs reviewed
    Verified --> [*]

Recovery does not patch the visible report first. It changes the governing source or claim record, then rebuilds normalized, review, admission, and publication descendants. This order prevents a map from disagreeing with the database that is supposed to explain it.

Recovery Changes A Decision Only Through Evidence¶

A recovery item is not complete because a blank field now has a value. The new material must close the specific missing edge that blocked the claim. Consider a project-context point awaiting sample recovery:

State	Required record
before recovery	project and paper identity, place statement, provisional context member, failed sample-evidence condition
recovered evidence	captured sample-bearing artifact with content identity and exact row locator
identity decision	project-owned stable sample key plus source-native labels and ambiguity disposition
claim review	sample-to-site, locality, chronology, and coordinate decisions evaluated independently
product effect	prior context member retained in history; replacement, qualification, or continued refusal explained at member level

flowchart LR
    Gap["named missing relation"] --> Artifact["captured supporting evidence"]
    Artifact --> Relation["governed object or relation"]
    Relation --> Review["new claim-specific decision"]
    Review --> Diff["member and accountability diff"]
    Value["plausible value without lineage"] -. "cannot close" .-> Gap

The publication count may remain unchanged even when the evidence class changes. Recovery review therefore compares stable identities, roles, precision, and qualifications—not totals alone.

Accept A Recovery¶

A recovery is complete when the intended evidence boundary is coherent and the causal diff is explainable:

the recovered artifact has stable source and content identity;
the object and claim owners are unambiguous or explicitly qualified;
original wording, nulls, conflicts, and precision remain preserved;
previous recovery and exclusion records close by reason, not disappearance;
every affected normalized and published member is identified;
unchanged descendants are also explainable when the new evidence does not cross their product contract.

Closing a recovery item does not delete its history. The prior state, recovered artifact, decision change, and affected descendants remain connected so a reader can explain both the earlier exclusion and the later admission.

Continue with record admission to interpret the revised decision, animal source intake for the literature and archive boundary, and verification evidence for claim-specific proof.