FIDEDocs
WorkspaceGraphGatewayFCP
Fide Context ProtocolGuidesSDKs
Evaluating

Statement Evaluation

Evaluation methods for assessing the factual accuracy and consensus of statements in the Fide Context Graph.

Fide-StatementAccuracy-v1

Determine factual validity based on evidence.

Is this specific statement factually true? This method is strictly for Objective Facts (e.g., "2 + 2 = 4", "Transaction 0x123 occurred").

Evaluation

Statement PartRaw IdentifierFide ID
SubjectThe Statement being evaluateddid:fide:0x60
PredicateFide-StatementAccuracy-v1 EvaluationMethod (e.g., GitHub spec link)did:fide:0xe5
ObjectVerdict value: -1, 0, or 1did:fide:0x66

Evaluation Process

  1. Input: Subject Statement + linked evidence statements (e.g., schema:citation references)
  2. Retrieval: Search external ground truth (Wikidata, Corporate DB) for the Subject
  3. Cross-Reference: Compare the Statement assertion against evidence
  4. Verdict Logic:
    • True (1): Supported by >1 trusted source or cryptographic proof
    • False (-1): Contradicted by >1 trusted source
    • Unverified (0): Insufficient data

Audit Trail: Link your verdict to the Action Entity that produced it using prov:wasGeneratedBy. This allows users to inspect the reasoning trace.

Fide-BridgingConsensus-v1

Measure agreement across diverse groups (Anti-Mob Rule).

Does this statement have broad consensus? This algorithm rewards agreement from opposing groups (Community Notes style) to reduce sybil-driven outcomes.

Evaluation

Statement PartRaw IdentifierFide ID
SubjectThe Statement being evaluateddid:fide:0x60
PredicateFide-BridgingConsensus-v1 EvaluationMethod (e.g., GitHub spec link)did:fide:0xe5
ObjectConsensus score: 0 to 100did:fide:0x66

Evaluation Process

  1. Input: Subject Statement + Set of Verdicts
  2. Cluster: Group voting Agents by historical voting similarity (e.g., matrix-factorization style clustering)
  3. Bridge: Identify statements that receive positive verdicts from opposing clusters
  4. Score: Higher scores go to statements that bridge the divide; lower scores go to statements only supported by one cluster

Output Scale

Score RangeMeaning
80-100Strong cross-group consensus
50-79Moderate agreement across some groups
20-49Limited consensus, mostly single-group support
0-19No bridging consensus detected

Entity Identity

Evaluation methods for determining whether two identifiers refer to the same real-world entity.

Statement Predicates

Evaluation methods for determining semantic relationships between predicates used in FCP statements.

On this page

Fide-StatementAccuracy-v1
Evaluation
Evaluation Process
Fide-BridgingConsensus-v1
Evaluation
Evaluation Process
Output Scale