id: https://nde.nl/ontology/hc/class/ConfidenceMethod
name: confidence_method_class
title: Confidence Method

prefixes:
  linkml: https://w3id.org/linkml/
  hc: https://nde.nl/ontology/hc/
  prov: http://www.w3.org/ns/prov#
  schema: http://schema.org/

imports:
  - linkml:types
  - ../slots/has_or_had_type
  - ../slots/has_or_had_description
  - ../slots/has_or_had_identifier

default_prefix: hc

classes:
  ConfidenceMethod:
    description: >-
      A method or algorithm used to calculate confidence scores.
      
      **USAGE**:
      Documents how confidence values were computed:
      - Fuzzy string matching algorithms
      - ML model predictions
      - Rule-based validation
      - XPath match verification
      - Human assessment
      
      **COMMON METHODS**:
      | Method | Description |
      |--------|-------------|
      | fuzzy_matching | Levenshtein, Jaro-Winkler, etc. |
      | xpath_validation | XPath match confidence |
      | llm_classification | LLM-based entity classification |
      | ml_prediction | Machine learning model output |
      | human_assessment | Manual quality assessment |
      | ensemble | Combined multiple methods |
      
    class_uri: prov:Plan
    
    exact_mappings:
      - prov:Plan
    
    close_mappings:
      - schema:HowTo
    
    slots:
      - has_or_had_type
      - has_or_had_description
      - has_or_had_identifier
    
    slot_usage:
      has_or_had_type:
        range: string
        required: true
        description: >-
          Type of confidence calculation method.
        examples:
          - value: "fuzzy_matching"
            description: String similarity algorithm
          - value: "ml_prediction"
            description: Machine learning model
          - value: "human_assessment"
            description: Manual human review
      
      has_or_had_description:
        range: string
        required: false
        description: >-
          Detailed description of the method implementation.
        examples:
          - value: "Jaro-Winkler similarity with 0.7 threshold"
            description: Fuzzy matching configuration
      
      has_or_had_identifier:
        range: string
        required: false
        description: >-
          Identifier for the method (e.g., algorithm name, model version).
        examples:
          - value: "rapidfuzz-2.15.1"
            description: Library version
    
    attributes:
      threshold:
        range: float
        description: >-
          Confidence threshold used by this method for accept/reject decisions.
        examples:
          - value: 0.85
            description: 85% threshold for acceptance
      
      is_deterministic:
        range: boolean
        description: >-
          Whether the method produces consistent results for same inputs.
    
    annotations:
      custodian_types: '["*"]'
      custodian_types_rationale: >-
        Confidence methods apply universally to data quality assessment.
      custodian_types_primary: "*"
      specificity_score: 0.25
      specificity_rationale: >-
        Low specificity - fundamental methodology documentation.
    
    examples:
      - value:
          has_or_had_type: "fuzzy_matching"
          has_or_had_description: "Levenshtein distance with ratio normalization"
          has_or_had_identifier: "rapidfuzz-levenshtein"
          threshold: 0.85
          is_deterministic: true
        description: Fuzzy string matching method
      
      - value:
          has_or_had_type: "llm_classification"
          has_or_had_description: "GPT-4 based entity type classification"
          has_or_had_identifier: "gpt-4-turbo-2024-04-09"
          is_deterministic: false
        description: LLM-based classification method
    
    comments:
      - Created from slot_fixes.yaml migration (2026-01-19)
      - Documents confidence calculation methodology
      - Used with ConfidenceScore class