core: type: Package publisher: worldview.genval.ai version: 1.0.0 imports: - publisher: kanonak.org packages: - package: core-rdf match: ^ version: 1.0.0 alias: rdfs - package: core-owl match: ^ version: 2.0.0 alias: owl - package: core-xsd match: ^ version: 1.0.0 alias: xsd # ============================================================================= # OVERVIEW # # This package defines the domain-general vocabulary for authoring falsifiable, # evidence-backed worldview snapshots under the Kanonak Protocol. A worldview # snapshot is a point-in-time record of "what we believe and why" - a list of # falsifiable theses, each with explicit invalidation conditions and supporting # evidence, plus an overarching narrative. # # The ontology has six concentric layers: # # 1. Snapshot / Thesis - the top-level units of authorship. # 2. Evidence base + four generic Evidence subclasses - the typed claims # that support theses. # 3. Status and Strength enumerations - small categorical vocabularies. # 4. Reference vocabulary stubs - abstract classes (Indicator, Event, # Institution, GovernmentBody, Location, Party, Role) whose concrete # instances live in domain-specific packages (e.g. worldview.genval.ai/ # finance for finance-domain instances). # 5. Condition DSL - a boolean condition tree (Condition + 11 concrete # subclasses arranged through three abstract intermediates: # NaryBoolCondition, ComparisonCondition, TimeBoundCondition) that # turns prose invalidation conditions into a structure a rules engine # can evaluate. # 6. Generic seed instances - small enumerations the DSL itself depends on # (Comparator, TimeUnit, Durability, EvaluationState) plus broad Policy # and Role members usable across domains. DerivedIndicator subclasses # (MovingAverage, etc.) are real classes rather than tagged-union # discriminators. # # Domain extensions (finance, climate, scientific, etc.) import this package, # subclass Evidence / Indicator / Event for their domain, and add named # instances of the abstract reference classes. # # This package is the canonical source of truth. Reference documentation # (markdown / HTML / JSON) is DERIVED from these `comment:` fields via # transformations - never authored separately. # ============================================================================= # ============================================================================= # 1. SNAPSHOT AND THESIS # ============================================================================= WorldviewSnapshot: type: rdfs.Class label: Worldview Snapshot comment: > A point-in-time record of a complete worldview. Conventionally a worldview package contains exactly ONE WorldviewSnapshot named `snapshot` plus a list of WorldviewThesis instances and the Evidence instances those theses cite. The snapshot's `observedAt` timestamp + the package's semver version together define the snapshot's identity. Snapshot package versions are immutable - a refresh writes a new version, never edits an existing one, so `git log` doubles as the audit trail of evolving belief. WorldviewThesis: type: rdfs.Class label: Worldview Thesis comment: > A single falsifiable claim within a worldview. Carries the prose statement (`thesisStatement`), an explicit machine-evaluable `invalidationCondition` that defines when the thesis breaks, a list of `supportingEvidence` URIs backing the claim, a calibrated `confidence` (0.0-1.0), and a `status` (active / retired / invalidated). A thesis without an invalidationCondition is unfalsifiable belief, not analysis. # --- WorldviewSnapshot properties --- observedAt: type: owl.DatatypeProperty domain: WorldviewSnapshot range: xsd.dateTime label: Observed At comment: > ISO-8601 timestamp marking the moment the snapshot was authored. Should reflect the data cutoff, not the file write time - a snapshot built from a market close should carry that close time. narrative: type: owl.DatatypeProperty domain: WorldviewSnapshot range: xsd.string label: Narrative comment: > A 1-3 paragraph human-readable summary of the regime, the day's pivotal events, and the rationale behind any confidence shifts since the prior snapshot. The narrative is for humans; the structured theses and evidence are for tools. hasThesis: type: owl.ObjectProperty domain: WorldviewSnapshot range: WorldviewThesis label: Has Thesis comment: > Ordered list of theses included in this snapshot. Order is significant only insofar as it affects the rendering order in transformations; the set is the source of truth, not the sequence. previousSnapshot: type: owl.ObjectProperty domain: WorldviewSnapshot range: WorldviewSnapshot label: Previous Snapshot comment: > Optional explicit reference to the WorldviewSnapshot this snapshot succeeds. The conventional package version chain (1.0.0 -> 1.0.1 -> 1.0.2) implicitly carries succession order, but this property makes it machine-walkable without inferring from version numbers and handles non-linear cases (a forked experimental branch that diverges and rejoins, or a regime-shift major version bump that skips intermediate patches). Time-series transformations should prefer this property when present; fall back to semver order when absent. If absent for a package's first snapshot, that's correct (no predecessor). # --- WorldviewThesis properties --- thesisStatement: type: owl.DatatypeProperty domain: WorldviewThesis range: xsd.string label: Thesis Statement comment: > Prose statement of the thesis. 2-4 sentences, falsifiable in spirit (the structured `invalidationCondition` makes "falsifiable" precise). Authored in plain language so a human reader can grasp the claim without parsing YAML. supportingEvidence: type: owl.ObjectProperty domain: WorldviewThesis range: Evidence label: Supporting Evidence comment: > Ordered list of references to Evidence instances (typically defined in the same snapshot package). Each cited Evidence should genuinely support the thesis - over-citing weak or tangential evidence weakens the thesis's credibility, not strengthens it. invalidationCondition: type: owl.ObjectProperty domain: WorldviewThesis range: Condition label: Invalidation Condition comment: > The structured machine-evaluable condition under which this thesis is invalidated. A `Condition` is a tagged-union expression tree (see Condition / ConditionForm below). When a rules engine evaluates the condition tree against current data and the engine returns `satisfied`, the thesis SHOULD be flagged for human review and a status flip to `cp.invalidated`. Partial / unsatisfied / unknown states are diagnostic inputs to thesis-confidence decisions but do not auto-mutate state. confidence: type: owl.DatatypeProperty domain: WorldviewThesis range: xsd.decimal label: Confidence comment: > Decimal in 0.0-1.0 expressing the author's conviction. Calibration scale (working heuristic, not a formal Brier-style target): 0.9-1.0 - thesis would survive any one of its invalidation conditions firing. 0.75-0.85 - high conviction; act on it; multiple independent supports. 0.5-0.7 - working hypothesis; size modestly. 0.3-0.45 - watching; one or two structural supports; invalidation in motion. 0.1-0.25 - mostly invalidated; consider retiring. 0.0 - should be `status: invalidated`, not active with low confidence. status: type: owl.ObjectProperty domain: WorldviewThesis range: ThesisStatus label: Status comment: > The thesis's lifecycle state. Transitions allowed - active to retired, active to invalidated. Not allowed - invalidated back to active (a falsified thesis stays falsified; if the same idea becomes valid again it deserves a new thesis with a new name and fresh framing). confidenceRationale: type: owl.DatatypeProperty domain: WorldviewThesis range: xsd.string label: Confidence Rationale comment: > Optional 1-2 sentence prose explaining why confidence shifted in this snapshot relative to the prior one. Splitting the rationale onto each thesis (vs. burying it in the snapshot's global narrative) lets a time-series view show "thesis X moved 0.5 -> 0.7 because $reason" per version without parsing free-form prose. Authors may leave this blank and use the snapshot's narrative for the same purpose; the property is for tooling that wants the rationale machine-attached to the confidence value. statusTransitionEvidence: type: owl.ObjectProperty domain: WorldviewThesis range: Evidence label: Status Transition Evidence comment: > Optional reference to the Evidence instance that triggered the most recent status transition (active -> retired, or active -> invalidated). Set when status changes; left absent on initial-active status. Time- series visualizations use this to annotate the timeline ("thesis invalidated by $evidence on $date") rather than parsing the invalidationCondition tree to figure out which condition fired. May be a list in v2 if multiple evidence trigger together; v1 single ref is the simpler case and covers most patterns. # ============================================================================= # 2. THESIS STATUS ENUM # ============================================================================= ThesisStatus: type: rdfs.Class label: Thesis Status comment: > Lifecycle state of a thesis. Closed enumeration - the three named instances below are the only valid values. active: type: ThesisStatus label: Active comment: > Thesis is in force, supporting evidence is current, invalidation condition has not fired. The default state for any thesis appearing in `hasThesis`. retired: type: ThesisStatus label: Retired comment: > Thesis is no longer in force but was NOT falsified - it lost relevance because the regime shifted, the framing became obsolete, or a follow-on thesis subsumed it. A retired thesis remains in the snapshot's `hasThesis` for history; consumers should not treat it as informing current allocation. invalidated: type: ThesisStatus label: Invalidated comment: > Thesis's invalidation condition fired and the thesis is now falsified. Stays in the snapshot's `hasThesis` permanently - the falsification record is part of the audit trail. A NEW thesis with a different framing should be authored if a related claim returns to relevance later. # ============================================================================= # 3. EVIDENCE BASE AND COMMON PROPERTIES # ============================================================================= Evidence: type: rdfs.Class label: Evidence comment: > Abstract base for any typed claim that supports or weakens a thesis. Never instantiated directly - authors instantiate one of the typed subclasses (PolicyEvidence, GeopoliticalEvidence, OfficialStatementEvidence, EconomicDataEvidence in this package; finance-specific subclasses in worldview.genval.ai/finance; future climate-specific subclasses in a climate package). Picking the most specific subclass is what lets future query tooling slice "show me all AnalystEvidence about gold targets in Q1" or "show me all GeopoliticalEvidence involving Iran in 2026". evidenceStatement: type: owl.DatatypeProperty domain: Evidence range: xsd.string label: Evidence Statement comment: > A 1-3 sentence prose claim distilled from the underlying source. NOT a paste of the full source. The statement is what the worldview reasons over; the `citation` URI is where to verify it. Keep statements specific and quantitative where possible. evidenceStrength: type: owl.ObjectProperty domain: Evidence range: EvidenceStrength label: Evidence Strength comment: > The author's assessment of how strongly this evidence supports the claim, derived primarily from source reputation and primacy (primary source vs. aggregator vs. anonymous). See EvidenceStrength named instances for the calibration heuristic. evidenceObservedAt: type: owl.DatatypeProperty domain: Evidence range: xsd.dateTime label: Evidence Observed At comment: > ISO-8601 timestamp of when the underlying fact was observed (price close time, news publication time, official meeting time, etc.). Optional but strongly preferred - a fact without a timestamp is hard to age out of relevance. citation: type: owl.DatatypeProperty domain: Evidence range: xsd.string label: Citation comment: > URI or other primary-source identifier (typically a URL) that a future auditor can follow to verify the recorded statement. The audit pointer. Optional but strongly preferred for evidence drawn from external sources. # ============================================================================= # 4. EVIDENCE STRENGTH ENUM # ============================================================================= EvidenceStrength: type: rdfs.Class label: Evidence Strength comment: > Categorical assessment of how strongly an Evidence instance supports its cited claim. Closed enumeration with three levels. evidence-strong: type: EvidenceStrength label: Strong comment: > Primary source, official press release, audited filing, or tier-1 wire (Reuters, Bloomberg, WSJ, FT, AP). Confirmed government action with named issuing body. The default level for evidence drawn directly from the institution that produced the fact. evidence-moderate: type: EvidenceStrength label: Moderate comment: > Tier-2 financial press (CNBC, MarketWatch, Yahoo Finance), industry trade press, or analyst notes without primary attribution. Suggestive but deserving of corroboration before driving high-confidence theses. evidence-weak: type: EvidenceStrength label: Weak comment: > Blog, secondary aggregation, anonymous source, or unverified social media. Should rarely raise thesis confidence on its own; useful as a leading indicator pending corroboration. # ============================================================================= # 5. GENERIC EVIDENCE SUBCLASSES # ============================================================================= PolicyEvidence: type: rdfs.Class subClassOf: Evidence label: Policy Evidence comment: > A government action - executive order, appointment, sanction, budget appropriation, regulatory ruling, treaty signing. Captures the policy as a discrete event rather than as a market reaction to it. GeopoliticalEvidence: type: rdfs.Class subClassOf: Evidence label: Geopolitical Evidence comment: > A geopolitical event - war, peace, military operation, election outcome, diplomatic accord. Distinct from PolicyEvidence in that it captures multi-party events between state-level actors rather than a single state's internal policy action. OfficialStatementEvidence: type: rdfs.Class subClassOf: Evidence label: Official Statement Evidence comment: > A direct quote or attributable statement from a named official - politician, executive, central banker, regulator, judge. The quote itself is the evidence; the statement's effects on markets are captured separately as MarketDataEvidence (in finance) or other domain-specific subclasses. EconomicDataEvidence: type: rdfs.Class subClassOf: Evidence label: Economic Data Evidence comment: > A macro economic data release - CPI, NFP, GDP, unemployment, treasury yield curve, breakeven inflation. Indicator-based rather than entity- based. The `indicatorName` should reference an Indicator instance (typically defined in a domain-specific package). # --- PolicyEvidence properties --- issuingBody: type: owl.ObjectProperty domain: PolicyEvidence range: GovernmentBody label: Issuing Body comment: > The government body or executive office that issued the policy. Reference to a GovernmentBody instance (typically defined in a domain-specific or country-specific package). policy: type: owl.ObjectProperty domain: PolicyEvidence range: Policy label: Policy Category comment: > Categorical type of policy action. References a Policy named instance. Generic categories (appointment, executive-order, budget, sanction, treaty, investigation) live in this package; domain-specific categories may be added in extension packages. effectiveDate: type: owl.DatatypeProperty domain: PolicyEvidence range: xsd.date label: Effective Date comment: > Date the policy took or takes effect. Distinct from announcement date when the two differ. Optional - some policies (public-pressure, fiscal-guidance) have no formal effective date. # --- GeopoliticalEvidence properties --- eventLocation: type: owl.ObjectProperty domain: GeopoliticalEvidence range: Location label: Event Location comment: > Reference to a Location instance describing where the event took place. Locations are typed entities, not free-form strings - use the Location class to enable cross-evidence queries like "all events in the Persian Gulf in 2026". partiesInvolved: type: owl.ObjectProperty domain: GeopoliticalEvidence range: Party label: Parties Involved comment: > Ordered list of references to Party instances representing actors in the event. Parties are typed entities (nations, organizations, individuals) rather than free-form strings. geopoliticalEventDate: type: owl.DatatypeProperty domain: GeopoliticalEvidence range: xsd.date label: Geopolitical Event Date comment: > Date the event occurred. For events with duration (campaigns, sieges, negotiations), use the start date and capture duration in the prose `evidenceStatement`. v2 may add structured duration. # --- OfficialStatementEvidence properties --- speaker: type: owl.ObjectProperty domain: OfficialStatementEvidence range: Party label: Speaker comment: > The named official making the statement. References a Party instance. Use a Party rather than a free-form string so the same speaker across multiple statements is unambiguously the same entity. speakerRole: type: owl.ObjectProperty domain: OfficialStatementEvidence range: Role label: Speaker Role comment: > The role the speaker held at the time of the statement. References a Role instance. Use Role rather than a free-form string to disambiguate "Fed Chair" vs "Fed Chair (outgoing)" vs "Fed Chair nominee" - those are distinct Role instances reflecting position in a transition lifecycle. statementVenue: type: owl.DatatypeProperty domain: OfficialStatementEvidence range: xsd.string label: Statement Venue comment: > Where the statement was delivered (press conference, hearing, written release, interview). String for v1; v2 may upgrade to a typed Venue class once usage patterns stabilize. statementDate: type: owl.DatatypeProperty domain: OfficialStatementEvidence range: xsd.date label: Statement Date comment: > Date of the statement. Distinct from `evidenceObservedAt` (the article publication time) when a statement is reported after the fact. # --- EconomicDataEvidence properties --- indicatorName: type: owl.ObjectProperty domain: EconomicDataEvidence range: Indicator label: Indicator comment: > The named Indicator this datapoint observes. References an Indicator instance defined in a domain package (e.g. `fin.10y-treasury-yield`, `fin.unemployment-rate`). Indicators are typed so multiple readings of the same indicator across snapshots are unambiguously comparable. indicatorValue: type: owl.DatatypeProperty domain: EconomicDataEvidence range: xsd.decimal label: Indicator Value comment: > The numeric reading. Units are carried by the Indicator instance, not the value - "4.42% 10y yield" stores `4.42` here and `pct` on the Indicator. Avoid stuffing units, qualifiers, or ranges into the value. For range readings, use evidenceStatement prose; v2 may add a structured range type. indicatorPeriod: type: owl.DatatypeProperty domain: EconomicDataEvidence range: xsd.string label: Indicator Period comment: > String describing the reading's reporting period (e.g. "April 2026", "Q1 2026", "trailing 25-30 years"). Free-form for v1; v2 may upgrade to a typed FiscalPeriod / RangeRef. # ============================================================================= # 6. REFERENCE VOCABULARY STUBS # # Abstract classes whose concrete instances are typically defined in # domain-specific packages. This package defines NO instances of these # classes - finance, climate, scientific extensions add their own. # ============================================================================= GovernmentBody: type: rdfs.Class label: Government Body comment: > A government body - executive office, legislature, regulatory agency, judicial body, sovereign wealth manager. Typically defined per-country in domain extensions (e.g. `gov-us.white-house`, `gov-us.dod`, `gov-us.fed`). The class itself is in core so PolicyEvidence can have a typed `issuingBody` regardless of country. Location: type: rdfs.Class label: Location comment: > A geographic or jurisdictional location - country, region, body of water, airspace, building, or named area. Locations are typed so the same place (e.g. "Persian Gulf") referenced across many GeopoliticalEvidence instances is unambiguously a single entity. Party: type: rdfs.Class label: Party comment: > A party in a geopolitical or institutional sense - nation, organization, or individual official. Used by GeopoliticalEvidence.partiesInvolved and OfficialStatementEvidence.speaker. Concrete Party instances live in domain packages. Role: type: rdfs.Class label: Role comment: > An office, title, or functional role held by a Party (typically an individual). Roles can be lifecycle-specific - "Fed Chair", "Fed Chair (outgoing)", and "Fed Chair nominee" are distinct Role instances during a transition because they carry different policy weight. Institution: type: rdfs.Class label: Institution comment: > A persistent organization - corporation, central bank, sell-side analyst house, multilateral body, NGO. Domain extensions subclass this for typed differentiation (CentralBank, AnalystInstitution, etc.). Indicator: type: rdfs.Class label: Indicator comment: > A named, repeatedly-observed numeric series - a market price, a yield, a macro statistic, an aggregate flow. Indicators are typed so cross- snapshot comparisons of "the same indicator" are unambiguous. Concrete Indicator instances live in domain packages; this package only declares the abstract class. Event: type: rdfs.Class label: Event comment: > A named category of event the runtime can detect against a curated event log. Used by Condition's event-condition kind. Concrete Event instances (e.g. `fin.evt-strait-of-hormuz-reopened`, `fin.evt-opec-supply-shock`) live in domain packages. Each named instance is a CATEGORY of event; specific event occurrences live in the runtime event log. # ============================================================================= # 7. POLICY-TYPE ENUMERATION (generic; finance and other domains can add more) # ============================================================================= Policy: type: rdfs.Class label: Policy comment: > A category of government action. Open enumeration - this package seeds widely-applicable categories; domain extensions can add specialized ones via additional named instances. Named instances represent KINDS of policy actions (executive-order, appointment, treaty), not specific policies; a specific policy action is a PolicyEvidence instance whose `policy` property points at the relevant Policy named instance. appointment: type: Policy label: Appointment comment: > Naming or confirming an individual to a government office or seat - cabinet appointments, judicial nominations, central bank chair nominations, board seats. executive-order: type: Policy label: Executive Order comment: > A directive issued by a head of state with the force of law within their constitutional authority. Captures both new orders and revocations. budget: type: Policy label: Budget comment: > Formal budget appropriation or budget request - whether enacted, vetoed, or in process. The `effectiveDate` carries the fiscal year start. sanction: type: Policy label: Sanction comment: > Economic or trade sanction imposed on a target party (nation, organization, individual). Sanctions removal is a separate Policy instance (`sanction-lifted`). sanction-lifted: type: Policy label: Sanction Lifted comment: > Removal or suspension of a previously-imposed sanction. treaty: type: Policy label: Treaty comment: > A formal international agreement - signed, ratified, withdrawn, or amended. Captures bilateral and multilateral instruments alike. investigation: type: Policy label: Investigation comment: > A formal investigative action - opening, halting, settling, or closing an inquiry. Investigations of officials are PolicyEvidence; the underlying conduct may also generate other Evidence types. regulatory-ruling: type: Policy label: Regulatory Ruling comment: > A formal ruling, order, or rulemaking by an executive-branch regulator. Distinct from `executive-order` (which is from the head of state). public-pressure: type: Policy label: Public Pressure comment: > A government actor publicly pressuring another institution for an outcome - executive criticizing the central bank, legislator demanding agency action. No formal force of law but materially shapes expectations. fiscal-pressure: type: Policy label: Fiscal Pressure comment: > A structural fiscal condition acting as a constraint on policy choices - e.g. a debt-service obligation that materially limits fiscal headroom. Captures the constraint, not a specific action. # ============================================================================= # 8. CONDITION DSL # # A boolean condition tree for thesis invalidation. Each form (And, Or, # Threshold, Event, Sequenced, ...) is a real subclass of Condition with # its own properties. Embedded Conditions inside `operands` / `after` / # `then` / etc. declare their concrete subclass via `type:` per the # kanonak-protocol@1.1.0 embedded-type-must-narrow-range rule. # ============================================================================= Condition: type: rdfs.Class label: Condition comment: > Abstract base for nodes in a boolean condition tree. Never instantiated directly - authors instantiate one of the subclasses below. A runtime evaluator walks the tree against current data plus an event log and returns an EvaluationState (satisfied, partially-satisfied, unsatisfied, unknown) plus a structured trace. # --- Boolean composition --- NaryBoolCondition: type: rdfs.Class subClassOf: Condition label: N-ary Boolean Condition comment: > Abstract intermediate for boolean conditions that compose N child conditions via a single connective (AND, OR). Captures the shared `operands` property so AndCondition and OrCondition don't redeclare it. AndCondition: type: rdfs.Class subClassOf: NaryBoolCondition label: And comment: > Boolean AND. Satisfied iff every operand is satisfied. Unsatisfied if any operand is unsatisfied. Unknown if any operand is unknown (and no operand is unsatisfied). Otherwise partially-satisfied. OrCondition: type: rdfs.Class subClassOf: NaryBoolCondition label: Or comment: > Boolean OR. Satisfied if any operand is satisfied. Unsatisfied iff every operand is unsatisfied. Unknown if any operand is unknown (and no operand is satisfied). Otherwise partially-satisfied. NotCondition: type: rdfs.Class subClassOf: Condition label: Not comment: > Boolean NOT. Inverts satisfied / unsatisfied of its single `operand`. Partial stays partial; unknown stays unknown. operands: type: owl.ObjectProperty domain: NaryBoolCondition range: Condition label: Operands comment: > Ordered list of child Conditions composed by the boolean connective. Used by AndCondition (all-must-pass) and OrCondition (any-passes). operand: type: owl.ObjectProperty domain: NotCondition range: Condition label: Operand comment: The single child Condition negated by NotCondition. # --- Comparison conditions --- ComparisonCondition: type: rdfs.Class subClassOf: Condition label: Comparison Condition comment: > Abstract intermediate for conditions that apply a Comparator to two sides plus an optional persistence window. Captures the shared `comparator`, `persistence`, `unit` properties used by both ThresholdCondition (right side is a literal value) and IndicatorComparison (right side is another Indicator). ThresholdCondition: type: rdfs.Class subClassOf: ComparisonCondition label: Threshold comment: > Numeric threshold check. Reads `observable`'s most recent value and compares to `threshold` using `comparator`. With `persistence`, the condition is partially-satisfied inside the persistence window and satisfied only after the window elapses with the comparator continuously true. IndicatorComparison: type: rdfs.Class subClassOf: ComparisonCondition label: Indicator Comparison comment: > Cross-indicator comparison. Compares `left` observable against `right` observable using `comparator`. Same persistence semantics as ThresholdCondition. Typical use: technical-analysis predicates ("SPY below 50d MA") where right is a DerivedIndicator on left. comparator: type: owl.ObjectProperty domain: ComparisonCondition range: Comparator label: Comparator comment: The comparison operator applied between the two sides. persistence: type: owl.ObjectProperty domain: ComparisonCondition range: PersistenceWindow label: Persistence comment: > Optional window during which the comparator must remain true before the condition is `satisfied`. Without a persistence window, a single satisfying reading flips the condition to satisfied; with one, the condition is `partially-satisfied` inside the window and `satisfied` only after the window elapses. unit: type: owl.DatatypeProperty domain: ComparisonCondition range: xsd.string label: Unit comment: > Free-form unit string documenting the threshold's units (e.g. "usd-per-bbl", "pct"). Informational; the underlying Indicator carries the canonical unit. v2 may upgrade to a typed Unit class. observable: type: owl.ObjectProperty domain: ThresholdCondition range: Indicator label: Observable comment: The Indicator whose value is compared against `threshold`. threshold: type: owl.DatatypeProperty domain: ThresholdCondition range: xsd.decimal label: Threshold comment: Numeric value compared against the observable. thresholdHigh: type: owl.DatatypeProperty domain: ThresholdCondition range: xsd.decimal label: Threshold High comment: > With comparator=between, the upper bound. Lower bound is `threshold`. The condition is satisfied when the observable lies in [threshold, thresholdHigh] inclusive. left: type: owl.ObjectProperty domain: IndicatorComparison range: Indicator label: Left comment: > The left-hand-side observable. May be either a base Indicator or a DerivedIndicator. right: type: owl.ObjectProperty domain: IndicatorComparison range: Indicator label: Right comment: > The right-hand-side observable. Typically a DerivedIndicator (e.g. a moving average of `left`) when the comparison expresses a technical-analysis condition. # --- Time-bound conditions (event-log queries + temporal sequencing) --- TimeBoundCondition: type: rdfs.Class subClassOf: Condition label: Time-Bound Condition comment: > Abstract intermediate for conditions that may be bounded by a `withinWindow` relative to evaluation time. Captures the shared `withinWindow` property. Concrete subclasses: EventCondition, SequencedCondition, IndicatorAvailable. EventCondition: type: rdfs.Class subClassOf: TimeBoundCondition label: Event comment: > Event-log presence check. Satisfied if an event matching `event` (with the optionally-specified `durability`) has been recorded in the event log within `withinWindow` if specified, or ever otherwise. The event log itself is runtime infrastructure outside this ontology. SequencedCondition: type: rdfs.Class subClassOf: TimeBoundCondition label: Sequenced comment: > Temporal sequencing. Satisfied when `after` first became satisfied at some time T0 AND `then` is satisfied when evaluated against time >= T0 (optionally within `withinWindow` of T0). Unknown until T0 is recorded. Used for "X following durable Y" patterns. IndicatorAvailable: type: rdfs.Class subClassOf: TimeBoundCondition label: Indicator Available comment: > Data-freshness guard. Satisfied if `indicator` has at least one observation within `withinWindow` (if specified). Used as a wrapper around ThresholdCondition leaves to prevent false-unsatisfied results when the underlying data feed is stale. withinWindow: type: owl.ObjectProperty domain: TimeBoundCondition range: TimeWindow label: Within Window comment: > Optional. Bounds the evaluation to a time window relative to evaluation time. For EventCondition - event must be recorded within. For SequencedCondition - then-condition must be satisfied within (relative to T0). For IndicatorAvailable - the indicator must have at least one observation within. event: type: owl.ObjectProperty domain: EventCondition range: Event label: Event comment: > The category of event the condition tests for, referencing an Event named instance. Resolved at evaluation time against a runtime-curated event log. durability: type: owl.ObjectProperty domain: EventCondition range: Durability label: Durability comment: > Optional. Requires the matched event to carry the specified durability classification. Filters the event log to events that have been judged durable - useful for "after durable peace" semantics where a transient ceasefire shouldn't satisfy the condition. after: type: owl.ObjectProperty domain: SequencedCondition range: Condition label: After comment: > The prerequisite Condition that must become satisfied at some time T0 before `then` is evaluated. The runtime tracks T0 so re-evaluation across days is consistent. then: type: owl.ObjectProperty domain: SequencedCondition range: Condition label: Then comment: > The Condition evaluated against time >= T0 (where T0 is when `after` first became satisfied). Optionally bounded by `withinWindow`. indicator: type: owl.ObjectProperty domain: IndicatorAvailable range: Indicator label: Indicator comment: > The Indicator whose data freshness is being asserted. Satisfied if the indicator has at least one observation within `withinWindow` (if specified) or any observation at all (if not). # --- Deadline check --- BeforeCondition: type: rdfs.Class subClassOf: Condition label: Before comment: > Deadline check. Satisfied if `before` becomes satisfied before `mustOccur` does - i.e. the deadline passed without the prerequisite materializing. The mirror of SequencedCondition for invalidations that fire on the absence of an expected event. before: type: owl.ObjectProperty domain: BeforeCondition range: Condition label: Before comment: > The deadline Condition - if `before` becomes satisfied before `mustOccur` does, BeforeCondition is satisfied (i.e. the deadline passed without the prerequisite occurring). mustOccur: type: owl.ObjectProperty domain: BeforeCondition range: Condition label: Must Occur comment: > The prerequisite that must become satisfied before `before` does, otherwise BeforeCondition fires. # --- Constants --- AlwaysSatisfied: type: rdfs.Class subClassOf: Condition label: Always Satisfied comment: > Constant. Always returns satisfied. Useful for testing condition trees and for placeholder leaves during incremental authoring. AlwaysUnsatisfied: type: rdfs.Class subClassOf: Condition label: Always Unsatisfied comment: > Constant. Always returns unsatisfied. Mirror of AlwaysSatisfied. # ============================================================================= # 9. COMPARATOR # ============================================================================= Comparator: type: rdfs.Class label: Comparator comment: > Numeric comparison operator. Closed enumeration - the six instances below are the v1 set. lt: type: Comparator label: Less Than comment: Strict less-than. Threshold value is exclusive. lte: type: Comparator label: Less Than or Equal comment: Less-than-or-equal. Threshold value is inclusive. gt: type: Comparator label: Greater Than comment: Strict greater-than. Threshold value is exclusive. gte: type: Comparator label: Greater Than or Equal comment: Greater-than-or-equal. Threshold value is inclusive. eq: type: Comparator label: Equal comment: > Numeric equality. Use sparingly with floats; prefer between with a small tolerance band when checking near-equality. between: type: Comparator label: Between comment: > Range membership. Requires both `threshold` (low bound) and `thresholdHigh` (high bound). Inclusive on both ends. # ============================================================================= # 10. TIME UNIT # ============================================================================= TimeUnit: type: rdfs.Class label: Time Unit comment: > Duration unit for PersistenceWindow and TimeWindow. Open enumeration - domain extensions can add specialized units (e.g. fomc-meetings in finance) where the abstract calendar units are insufficient. minutes: type: TimeUnit label: Minutes comment: One minute = 60 seconds. hours: type: TimeUnit label: Hours comment: One hour = 60 minutes. trading-days: type: TimeUnit label: Trading Days comment: > Days the relevant market is open. Implementation depends on the Indicator's market calendar; the runtime is responsible for resolving "5 trading days" against the appropriate calendar. calendar-days: type: TimeUnit label: Calendar Days comment: 24-hour calendar days, regardless of market or business calendar. months: type: TimeUnit label: Months comment: > Calendar months. "3 months" means three calendar-month boundaries have been crossed since the start of the window. fiscal-quarters: type: TimeUnit label: Fiscal Quarters comment: > Standard quarterly periods (Q1, Q2, Q3, Q4). For entities on non-calendar fiscal years, the runtime should resolve against the relevant entity's fiscal calendar. # ============================================================================= # 11. DURABILITY # ============================================================================= Durability: type: rdfs.Class label: Durability comment: > Classification of how long an event's effects are expected to last. Used by event-condition to filter the event log. The judgment is the event-log curator's, not auto-derived. durable: type: Durability label: Durable comment: > The event's effects persist over a meaningful timeframe. Examples - a signed treaty, a confirmed appointment, a structurally-implemented sanction, a regulatory rulemaking that takes effect. transient: type: Durability label: Transient comment: > The event has occurred but its effects are not expected to persist. Examples - a verbal pause in operations, a preliminary indication of intent, a non-binding statement of position. unspecified: type: Durability label: Unspecified comment: > The curator has not yet judged durability. Default for newly-recorded events. event-condition with durability=unspecified matches any event of the type regardless of durability classification. # ============================================================================= # 12. EVALUATION STATE # ============================================================================= EvaluationState: type: rdfs.Class label: Evaluation State comment: > The four possible outcomes of evaluating a Condition tree. Carried in ConditionResult instances at runtime. Closed enumeration. satisfied: type: EvaluationState label: Satisfied comment: > The condition is fully satisfied against current data. For an invalidationCondition root, this signals the thesis SHOULD be flagged for human review and a status flip to invalidated. partially-satisfied: type: EvaluationState label: Partially Satisfied comment: > The condition is not yet satisfied but is making progress toward satisfaction (e.g. inside a persistence window with the underlying comparator currently true; sequenced-condition with `after` satisfied but `then` not yet). A leading indicator that the thesis may invalidate. unsatisfied: type: EvaluationState label: Unsatisfied comment: > The condition is not satisfied and not currently making progress toward satisfaction. The thesis remains intact under this dimension. unknown: type: EvaluationState label: Unknown comment: > The condition cannot be evaluated due to missing or stale data, or an unresolved event-log query. Distinct from unsatisfied - "we don't know" is operationally important to surface rather than silently treating as unsatisfied. # ============================================================================= # 13. PERSISTENCE WINDOW AND TIME WINDOW # ============================================================================= PersistenceWindow: type: rdfs.Class label: Persistence Window comment: > A duration over which a condition's underlying comparator must remain continuously true before the condition is satisfied. Composes a `duration` (integer) with a TimeUnit reference. Authors using this package must reference the `duration` property with the package alias (e.g. `wv.duration: 30`) because the bare name collides with `xsd:duration` (the built-in XSD datatype). duration: type: owl.DatatypeProperty domain: PersistenceWindow range: xsd.integer label: Duration comment: > Number of TimeUnit-units to persist. Positive integer. Must be referenced as `wv.duration` (or whatever alias your import uses) due to the bare-name collision with `xsd:duration`. windowUnit: type: owl.ObjectProperty domain: PersistenceWindow range: TimeUnit label: Window Unit comment: > The TimeUnit that scales `duration`. Together they define the window size (e.g. duration=30, windowUnit=calendar-days = 30 calendar days). TimeWindow: type: rdfs.Class label: Time Window comment: > A bounded duration relative to evaluation time, used by event-condition and sequenced-condition to constrain "within the past N units" matching. Composes a `windowDuration` with a `windowTimeUnit`. windowDuration: type: owl.DatatypeProperty domain: TimeWindow range: xsd.integer label: Window Duration comment: Number of TimeUnit-units the window spans. windowTimeUnit: type: owl.ObjectProperty domain: TimeWindow range: TimeUnit label: Window Time Unit comment: The TimeUnit that scales `windowDuration`. # ============================================================================= # 14. DERIVED INDICATOR # ============================================================================= DerivedIndicator: type: rdfs.Class subClassOf: Indicator label: Derived Indicator comment: > Abstract subclass of Indicator for indicators whose values are computed at evaluation time from another Indicator. Distinct from a base Indicator whose values come from a data feed. Concrete subclasses (currently MovingAverage; future RollingStdev, PercentChange, ZScore) capture the specific derivation logic. Embedded DerivedIndicator instances declare their concrete subclass via `type:` per the kanonak-protocol@1.1.0 embedded-type-must-narrow-range rule. MovingAverage: type: rdfs.Class subClassOf: DerivedIndicator label: Moving Average comment: > Simple moving average of the source Indicator over `period` units of `periodUnit`. The runtime computes by averaging the trailing `period` observations. of: type: owl.ObjectProperty domain: DerivedIndicator range: Indicator label: Of comment: The source Indicator the derivation operates on. period: type: owl.DatatypeProperty domain: DerivedIndicator range: xsd.integer label: Period comment: > The lookback period for the derivation, expressed in TimeUnit-units of the periodUnit. For MovingAverage, the number of bars in the average (50, 200, etc.). periodUnit: type: owl.ObjectProperty domain: DerivedIndicator range: TimeUnit label: Period Unit comment: TimeUnit scaling `period`. Most commonly `trading-days`. # ============================================================================= # 15. ROLE - generic seed instances # # A few broadly-applicable Role instances any worldview can reference. # Domain extensions add specialized roles (head-of-state-of-us, central- # bank-chair, etc.). # ============================================================================= head-of-state: type: Role label: Head of State comment: > Generic role for the chief executive of a sovereign nation. Specialize per country in domain extensions when the role-name disambiguation matters (President-of-US vs Prime-Minister-of-UK). legislator: type: Role label: Legislator comment: Generic role for an elected member of a legislative body. regulator: type: Role label: Regulator comment: > Generic role for an executive-branch regulator at the agency or director level. central-banker: type: Role label: Central Banker comment: > Generic role for a member of a central bank's policymaking committee. Specialize in finance extensions when chair vs governor vs regional- president disambiguation matters. judge: type: Role label: Judge comment: Generic role for a judicial officer. journalist: type: Role label: Journalist comment: > Generic role for an attributable financial-press author. Used when the press is the speaker rather than the venue.