Reputation / Trust Assessment
Describe one purpose-bound assessment of a person or party, with issuer accountability, evidence, qualified conclusions and a contestable revision lifecycle.
Bundle → Layer → Finding → Questions Filled
3 bundles · 3 layers · 5 findings · 10 questions
Scope and issuer Who assesses whom and for what.
Assessment frame
Subject, purpose and issuer.
Purpose and scope
The purpose and context for which trust is assessed.
- For which purpose and context is the trust assessment made?
- Is its use limited to that purpose, and by what rule?
Issuer and method
The issuing party and the method it applied.
- Who issued the assessment, and under what published method?
- Was the assessment automated, human or both?
Evidence and result What the judgement rests on.
Evidence base
The facts and signals used.
Evidence items
The records used, with dates and sources.
- Which records and signals were used, and how old are they?
- Were any sensitive characteristics used directly or by proxy?
Result and validity
The score or rating and how long it holds.
- What rating or level was given, and on what scale?
- When does the assessment expire or need renewal?
Contest and correction How the subject can challenge it.
Appeal lifecycle
Notice, dispute and correction.
Notice and appeal
Whether the subject was informed and can contest.
- Was the subject told about the assessment and its main reasons?
- Which appeals or corrections have been filed, and with what outcome?
Classifiers Filled
- Family
- World Models
- Category
- Society, people and institutions
- Entry kind
- entity
- Navigation path
- NAV.SOC.PER.TRU
- Domain
- SOC.PER.TRU
- Industry
- Cross-industry
- Tags
- reputationtrustassessmentsoc.per.tru
What it is Filled
A reputation or trust assessment is a recorded judgement by an identified issuer about how far a person or party can be trusted for a stated purpose, based on evidence such as transactions, reviews, verified credentials or conduct, with a lifecycle that allows correction and appeal. It is a scoped, evidence-backed assessment, not a general social score, and differs from a credential, which attests a fact rather than rating trust.
In scope
- Assessment identity, issuer and subject references, intended reliance context and accountable stewardship
- Evidence selection and provenance, method and scale definitions, uncertainty and bounded result assertions
- Validity, restricted use, explanation, contest, correction, supersession and controlled record continuity
Out of scope
- Person and organization masters, individual review publication platforms and credential issuance systems
- General social scoring, protected-trait inference, surveillance, universal trust rankings and automated adverse decisions
- Credit reporting, employment, housing or other regulated eligibility determinations, dispute adjudication and runtime enforcement engines; these require separately governed profiles and masters
Why it exists Filled
Describe one purpose-bound assessment of a person or party, with issuer accountability, evidence, qualified conclusions and a contestable revision lifecycle.
Distinguishing features Filled
- Scoped to a stated purpose and issuer, unlike a general social score.
- A judgement built on evidence, unlike a credential or attestation of a single fact.
- Has a lifecycle with expiry, correction and appeal.
- Distinct from the person or party it is about and from individual reviews that feed it.
What robots and AI may and may not do Filled
Must not
- Build general-purpose social scores of people.
- Use protected characteristics or their proxies to rate trust.
- Make legally or similarly significant decisions about a person solely from an automated score.
- Reuse an assessment for a purpose its issuer did not state.
- Publish a person's assessment beyond those entitled to see it.
Only with a human decision
- Decisions with legal or similarly significant effect on the subject.
- Resolving an appeal against an assessment.
May
- Retrieve an assessment with its issuer, purpose, method and date.
- Explain the main factors behind an assessment to the subject.
- Record disputes and route them to the issuer.
- Flag assessments that are expired or used outside their stated purpose.
Moral aspects Filled
- Trust scores can lock people out of work, housing, credit or platforms, often without them knowing.
- Biased evidence or methods reproduce discrimination at scale.
- Fake reviews and manipulation can destroy or inflate reputations unfairly.
Who is affected
- People and parties being assessed
- Parties relying on assessments
- Issuers and platforms
Owners Filled
Steward
The issuer that defines the method, makes the assessment and handles appeals.
Roles
- Assessment custodian
- Maintain assessment identity, authorized records and corrections
- Issuer or delegated assessor
- Support contextual assertions with method, evidence and mandate
- Subject or identity steward
- Resolve identity and submit authorized challenges without unrestricted editing of issuer claims
- Reviewer
- Assess challenges and method limitations within an evidenced mandate and conflict policy
- Access and records officer
- Control disclosure, retention, restrictions and disposal
- Relying party
- Check purpose and limitations; hold separate responsibility for any use decision
Master systems
- Reputation or review platform database
- Risk and trust decisioning system
Links to other meta-models Filled
references
- WM-PER-001 - Candidate person master binding derived from registry parent. No inheritance or ownership transfer; verify and pin before integration.
- External review and transaction masters - Input observations, moderation outcomes and transaction evidence remain in their masters. This model stores qualified references only.
- External decision and dispute case masters - Reference reliance decisions and competent review outcomes without owning eligibility, dispute adjudication or enforcement.
aligned
- Verifiable Credentials Data Model v2.0 - Optional representation of issuer claims; explicit profile needed. Credential verification does not establish assessment truth or fitness for purpose.
- PROV-O - Attribution, derivation and revision mapping candidate, not a complete runtime provenance or audit implementation.
neighbor
- WM-PER-001 registry parent - Candidate subject reference only. An assessment is about a person; it is not a person subtype and does not inherit identity stewardship authority. Person or explicitly authorized identity steward controls identity corrections; an authorized assessment custodian manages the assessment record.
- Credential and attestation records - A credential can encode evaluative claims. Credential format, proof and issuer identity are separate from the assessment meaning and the relying party decision to trust it.
- Reviews and transactions - Referenced observations or opinions are inputs. Their collection, moderation, publication and transaction lifecycle remain externally mastered; aggregation is not automatic corroboration.
- Decision and appeal case masters - The assessment records reliance restrictions and references to competent review outcomes. It does not make an eligibility decision, resolve a legal dispute or operate the receiving decision system.
parent
- WM-PER-001
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- Identified by the issuer's assessment identifier, the subject's identifier and the assessment date.
- Verifiable forms may be expressed as signed credentials with a credential identifier.
Direct properties not applicable Not applicable
Not applicable
The subject is an institutional or informational record with no physical quantities of its own.
Recognition optional Filled
- Marked by a subject, an issuer, a purpose, a rating and a date.
- Often confused with a credential, an individual review or a credit report.
Capabilities and actions required Filled
- Can be issued, updated, expired, disputed and corrected.
- Can be shared with relying parties under the subject's consent or a legal basis.
Hazards and failure modes required Filled
- Unfair exclusion from services through wrong or stale assessments.
- Manipulation through fake reviews or sybil accounts.
Standards and interfaces required Filled
- W3C Verifiable Credentials Data Model for portable signed assessments.
- ISO 20488 for online consumer reviews.
- Data subject access and objection procedures under data protection law.
Context of use required Filled
- Used on marketplaces, sharing platforms, supplier vetting and online communities.
- Automated decisions and certain scoring uses are restricted by data protection and AI law.
Sources Filled
- Verifiable Credentials Data Model v2.0 - World Wide Web Consortium
- PROV-O: The PROV Ontology - World Wide Web Consortium
- Artificial Intelligence Risk Management Framework (AI RMF 1.0) - National Institute of Standards and Technology
- The Consumer Reviews and Testimonials Rule: Questions and Answers - Federal Trade Commission
- Regulation (EU) 2016/679 - General Data Protection Regulation - European Union
- Regulation (EU) 2024/1689 - Artificial Intelligence Act - European Union
- ISO 20488:2018 - Online consumer reviews - Principles and requirements for their collection, moderation and publication - International Organization for Standardization
- Regulation (EU) 2016/679 General Data Protection Regulation, Article 22, European Union
- Regulation (EU) 2024/1689 Artificial Intelligence Act, European Union
- ISO 20488 Online consumer reviews - Principles and requirements, International Organization for Standardization
Open questions
- Verify full current source texts, effective versions, licensing, sector applicability and applicable legal interpretations, including assessment production that materially influences decisions.
- Develop nested schemas and pinned mappings with acceptance fixtures for ambiguity, missing evidence, false signed claims, duplicated inputs, stale status, context mismatch, contested results, recipient correction and lawful erasure.
- Restore independent external review before any canonical or publishable-draft promotion.
- Independent external review is absent; a separate local self-audit is not an independent provider.
- Direct HTTP checks are deferred to the coordinator under the owner-reported sandbox restriction. EU full texts and current consolidated versions, W3C errata and ISO normative clauses are not fully verified.
- Sector-specific credit, employment, housing and other eligibility profiles, non-EU privacy regimes and case law require separate research.
- Nested instance schemas, method validation, quantitative thresholds, executable mappings, runtime bindings and acceptance fixtures are not implemented.
Machine files
Provenance
world-models research · reviewable-draft
Built from: models/wm-per-015-reputation-trust-assessment/spec.yaml, ver-cy/world-models/card-supplements/wm-per-015-reputation-trust-assessment.json