05 / OPEN KNOWLEDGE INFRASTRUCTURE

From isolated claims to a traceable evidence network.

GEO—the Global Evidence Observatory—is a proposed open knowledge infrastructure for connecting sources, evidence, claims, models, reviews, corrections, versions, and disputes.

What is GEO?

GEO is a proposed research infrastructure for recording how knowledge claims are produced, supported, challenged, modeled, reviewed, corrected, and superseded. It is designed for researchers, readers, institutions, and computational agents that need to inspect the full evidence chain rather than consume an isolated conclusion.

GEO is not a truth engine, an authority registry, or a claim that all civilization knowledge has been verified. Its purpose is to preserve traceability, uncertainty, disagreement, and revision.

Trace claims to evidence. Keep correction possible.

Why civilization research needs an evidence network

Civilization-scale questions cross disciplines, regions, languages, institutions, and long time horizons. Conventional publication formats often preserve conclusions better than the relations needed to audit them.

Fragmented evidence

Historical records, datasets, field studies, simulations, and policy documents remain separated across repositories and disciplines.

Lost provenance

A statistic or quotation may travel far from its original context, collection method, limitations, and version.

Detached models

Models can appear precise while their assumptions, input evidence, transformations, and failed tests remain difficult to inspect.

Invisible disagreement

Criticism, replication failures, minority interpretations, corrections, and unresolved disputes are rarely linked to the original claim.

Source–Evidence–Claim–Model–Review

The core chain separates five research objects so that a conclusion can be examined at every step.

SOURCE

Original dataset, archive, document, interview, observation, or other record, with ownership, access, and context retained.

EVIDENCE

A bounded extract, observation, or measurement derived from a source and accompanied by method, uncertainty, and scope.

CLAIM

An explicit, scoped, and testable proposition stating what is asserted, for whom, where, when, and with what confidence.

MODEL

A formal, statistical, causal, or computational representation that links claims and evidence through declared assumptions.

REVIEW

Independent assessment, replication, criticism, ethical evaluation, or editorial decision, with reviewer role and outcome recorded.

Knowledge graph

The graph records research objects and typed relations. It makes dependencies visible; it does not convert relationships into certified truth.

Proposed node types

SourceEvidenceDatasetClaimIndicatorModelReviewVersionDispute

Proposed relation types

derived_from: Evidence → Source
supports: Evidence → Claim
contradicts: Evidence → Claim
uses: Model → Dataset / Evidence
tests: Model / Study → Claim
reviews: Review → Claim / Model
supersedes: Version → Version
corrects / disputes: Correction or Dispute → Research object

Claim states

01Proposed02Source-linked03Supported04Contested05Under review06Replicated07Revised08Withdrawn

Evidence provenance

Every evidence object should retain enough information for another researcher to find the source, understand the transformation, and evaluate limits.

Persistent identifier

Stable URI or record ID

Creator or custodian

Who created, collected, or maintains the source

Collection context

Method, place, date, population, and coverage

Access and rights

License, consent, restrictions, and access date

Transformation chain

Extraction, cleaning, coding, normalization, and derived variables

Integrity record

File version, timestamp, and checksum where appropriate

Uncertainty

Measurement error, confidence range, missingness, and ambiguity

Bias and limitations

Known selection effects, exclusions, conflicts, and scope limits

Citation path

Human-readable citation and resolvable source link

Version governance and dispute records

Correction must be visible without erasing the history needed to understand how the knowledge changed.

Version governance

01
Released records are immutable; changes create a new linked version.
02
Major, minor, and correction releases describe what changed and why.
03
Superseded records remain accessible and point to their successors.
04
Withdrawals record reason, authority, date, and affected dependencies.

Dispute record

01
Status: open, under review, resolved, unresolved, or withdrawn.
02
Record the disputed object, competing positions, evidence, reviewers, and decision rationale.
03
Preserve legitimate minority interpretations and unresolved uncertainty.
04
No silent deletion, retroactive “verification,” or removal of criticism.

Machine-readable data

The prototype will define open schemas before it defines an interface. Proposed formats remain subject to technical and ethical review.

JSON / JSON-LD

Primary structured records, linked identifiers, language tags, and semantic relations.

CSV / Parquet

Tabular observations, indicators, and reproducible analytical inputs.

RDF export

Optional graph interchange for compatible research systems.

Schema and API

Documented object definitions, validation rules, query endpoints, and versioned exports.

Trust metadata

Timestamps, checksums, licenses, provenance links, uncertainty fields, and review states.

Current state: Concept Prototype V0.1

The public page documents an intended architecture. It does not imply that a production graph, global corpus, or review institution is operating.

Defined in V0.1

Core object model and relation vocabulary
Five-stage evidence chain
Initial provenance requirements
Version, correction, withdrawal, and dispute principles
Human-readable three-language concept interface

Not yet operational

Production knowledge-graph database
Public ingestion and query API
Automated extraction or verification pipeline
Validated global source corpus
Independent governance and review body

Next-stage prototype plan

The next phase should turn the concept into a bounded, auditable research prototype before any claim of global coverage.

Schema and identifiers

Publish object definitions, relation rules, proposition states, multilingual fields, and persistent identifier policy.

Graph store and ingestion

Build a minimal repository, manual submission workflow, validation checks, and reproducible import process.

Seed evidence corpus

Test the schema on a small set of documented civilization-research cases with competing claims and mixed source types.

Governance workflow

Implement review, correction, version, withdrawal, dispute, ethics, privacy, and access-control procedures.

Open interface and evaluation

Release documented exports and a read-only API prototype; conduct independent usability, security, provenance, and bias review.

10 / EPISTEMIC BOUNDARY

GEO should make the path from source to conclusion inspectable. A graph can expose relations and revision history; it cannot by itself determine truth, legitimacy, or policy.