Class: AIModel

Trained or rule-based model embedded in an AI system. Carries paradigm, algorithm family, dataset references and version metadata.

URI: iso22989:AIModel

 classDiagram
    class AIModel
    click AIModel href "../AIModel/"
      NamedEntity <|-- AIModel
        click NamedEntity href "../NamedEntity/"


      AIModel <|-- NeuralNetworkModel
        click NeuralNetworkModel href "../NeuralNetworkModel/"


      AIModel : activation_function





        AIModel --> "0..1" ActivationFunctionType : activation_function
        click ActivationFunctionType href "../ActivationFunctionType/"



      AIModel : algorithm_family





        AIModel --> "0..1" MLAlgorithmFamily : algorithm_family
        click MLAlgorithmFamily href "../MLAlgorithmFamily/"



      AIModel : aliases

      AIModel : catastrophic_forgetting_risk

      AIModel : clause_reference

      AIModel : description

      AIModel : engineering_approach





        AIModel --> "0..1" EngineeringApproach : engineering_approach
        click EngineeringApproach href "../EngineeringApproach/"



      AIModel : hyperparameters

      AIModel : id

      AIModel : inference_latency_ms

      AIModel : model_compression_applied

      AIModel : model_paradigm





        AIModel --> "0..1" MachineLearningParadigm : model_paradigm
        click MachineLearningParadigm href "../MachineLearningParadigm/"



      AIModel : model_version

      AIModel : name

      AIModel : neural_network_architecture





        AIModel --> "0..1" NeuralNetworkArchitecture : neural_network_architecture
        click NeuralNetworkArchitecture href "../NeuralNetworkArchitecture/"



      AIModel : parameter_count

      AIModel : preferred_label

      AIModel : see_also_uri

      AIModel : supports_continuous_learning

      AIModel : test_dataset





        AIModel --> "0..1" Dataset : test_dataset
        click Dataset href "../Dataset/"



      AIModel : trained_on

      AIModel : training_dataset





        AIModel --> "0..1" Dataset : training_dataset
        click Dataset href "../Dataset/"



      AIModel : training_duration

      AIModel : training_phenomena





        AIModel --> "*" NeuralNetworkPhenomenon : training_phenomena
        click NeuralNetworkPhenomenon href "../NeuralNetworkPhenomenon/"



      AIModel : validation_dataset





        AIModel --> "0..1" Dataset : validation_dataset
        click Dataset href "../Dataset/"

Inheritance

NamedEntity
- AIModel
  - NeuralNetworkModel

Slots

Name	Cardinality and Range	Description	Inheritance
model_paradigm	0..1 MachineLearningParadigm	Machine-learning paradigm under which the model was trained	direct
algorithm_family	0..1 MLAlgorithmFamily	Algorithm family the model belongs to	direct
engineering_approach	0..1 EngineeringApproach	Non-learning engineering approach used (for symbolic/knowledge-based models)	direct
training_dataset	0..1 Dataset	Dataset used to fit the model	direct
validation_dataset	0..1 Dataset	Dataset used for hyperparameter tuning and model selection	direct
test_dataset	0..1 Dataset	Dataset used to estimate generalisation performance	direct
hyperparameters	* String	Free-form record of model hyperparameter settings	direct
model_version	0..1 String	Version identifier of the trained model artefact	direct
trained_on	0..1 String	Date or version reference for when the model was last trained	direct
supports_continuous_learning	0..1 Boolean	Whether a model supports continuous or online learning (Clause 3	direct
catastrophic_forgetting_risk	0..1 ConfidenceScore	Estimated risk of catastrophic forgetting on retraining (Clause 5	direct
parameter_count	0..1 Integer	Approximate count of trainable model parameters	direct
training_duration	0..1 DurationType	Wall-clock duration of model training, expressed as an ISO 8601 duration	direct
inference_latency_ms	0..1 Float	Typical end-to-end inference latency in milliseconds	direct
model_compression_applied	0..1 Boolean	Whether the model has had compression or distillation applied for deployment ...	direct
neural_network_architecture	0..1 NeuralNetworkArchitecture	Neural-network architecture, when algorithm_family is neural_network (Clauses...	direct
activation_function	0..1 ActivationFunctionType	Predominant activation function used in the network (Clause 3	direct
training_phenomena	* NeuralNetworkPhenomenon	Training-time phenomena observed for the model (Clause 3	direct
id	1 Uriorcurie	Unique CURIE or URI identifying the entity	NamedEntity
name	1 String	Human-readable label for the entity	NamedEntity
description	0..1 String	Free-text description of the entity (paraphrased; verbatim ISO text excluded)	NamedEntity
clause_reference	0..1 String	ISO/IEC 22989:2022 clause identifier (e	NamedEntity
aliases	* String	Alternative names or synonyms for the term	NamedEntity
preferred_label	0..1 String	Preferred natural-language label	NamedEntity
see_also_uri	* Uriorcurie	Pointers to related external resources or term records	NamedEntity

Usages

used by	used in	type	used
AISystem	models	range	AIModel
Prediction	produced_by	range	AIModel
InferenceEngine	uses_model	range	AIModel
AIConceptsCollection	ai_models	range	AIModel
CatastrophicForgetting	affected_model	range	AIModel
Inference	over_model	range	AIModel

In Subsets

Identifier and Mapping Information

Schema Source

from schema: https://w3id.org/lmodel/iso22989

Mappings

Mapping Type	Mapped Value
self	iso22989:AIModel
native	iso22989:AIModel
related	nist_ai_rmf:Model
close	nist_ai_100_1:AiSystemDimensionEnum#AI_MODEL

LinkML Source

Direct

name: AIModel
description: Trained or rule-based model embedded in an AI system. Carries paradigm,
  algorithm family, dataset references and version metadata.
in_subset:
- terminology
- ai_concepts
from_schema: https://w3id.org/lmodel/iso22989
close_mappings:
- nist_ai_100_1:AiSystemDimensionEnum#AI_MODEL
related_mappings:
- nist_ai_rmf:Model
is_a: NamedEntity
slots:
- model_paradigm
- algorithm_family
- engineering_approach
- training_dataset
- validation_dataset
- test_dataset
- hyperparameters
- model_version
- trained_on
- supports_continuous_learning
- catastrophic_forgetting_risk
- parameter_count
- training_duration
- inference_latency_ms
- model_compression_applied
attributes:
  neural_network_architecture:
    name: neural_network_architecture
    description: Neural-network architecture, when algorithm_family is neural_network
      (Clauses 3.4, 5.12.1).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    domain_of:
    - AIModel
    range: NeuralNetworkArchitecture
  activation_function:
    name: activation_function
    description: Predominant activation function used in the network (Clause 3.4.1).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    domain_of:
    - AIModel
    - Neuron
    range: ActivationFunctionType
  training_phenomena:
    name: training_phenomena
    description: Training-time phenomena observed for the model (Clause 3.4).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    domain_of:
    - AIModel
    range: NeuralNetworkPhenomenon
    multivalued: true

Induced

name: AIModel
description: Trained or rule-based model embedded in an AI system. Carries paradigm,
  algorithm family, dataset references and version metadata.
in_subset:
- terminology
- ai_concepts
from_schema: https://w3id.org/lmodel/iso22989
close_mappings:
- nist_ai_100_1:AiSystemDimensionEnum#AI_MODEL
related_mappings:
- nist_ai_rmf:Model
is_a: NamedEntity
attributes:
  neural_network_architecture:
    name: neural_network_architecture
    description: Neural-network architecture, when algorithm_family is neural_network
      (Clauses 3.4, 5.12.1).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: NeuralNetworkArchitecture
  activation_function:
    name: activation_function
    description: Predominant activation function used in the network (Clause 3.4.1).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    - Neuron
    range: ActivationFunctionType
  training_phenomena:
    name: training_phenomena
    description: Training-time phenomena observed for the model (Clause 3.4).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: NeuralNetworkPhenomenon
    multivalued: true
  model_paradigm:
    name: model_paradigm
    description: Machine-learning paradigm under which the model was trained.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: MachineLearningParadigm
  algorithm_family:
    name: algorithm_family
    description: Algorithm family the model belongs to.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: MLAlgorithmFamily
  engineering_approach:
    name: engineering_approach
    description: Non-learning engineering approach used (for symbolic/knowledge-based
      models).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: EngineeringApproach
  training_dataset:
    name: training_dataset
    description: Dataset used to fit the model.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: Dataset
  validation_dataset:
    name: validation_dataset
    description: Dataset used for hyperparameter tuning and model selection.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: Dataset
  test_dataset:
    name: test_dataset
    description: Dataset used to estimate generalisation performance.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: Dataset
  hyperparameters:
    name: hyperparameters
    description: Free-form record of model hyperparameter settings.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: string
    multivalued: true
  model_version:
    name: model_version
    description: Version identifier of the trained model artefact.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: string
  trained_on:
    name: trained_on
    description: Date or version reference for when the model was last trained.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: string
  supports_continuous_learning:
    name: supports_continuous_learning
    description: Whether a model supports continuous or online learning (Clause 3.1.9,
      5.11.9.2).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: boolean
  catastrophic_forgetting_risk:
    name: catastrophic_forgetting_risk
    description: Estimated risk of catastrophic forgetting on retraining (Clause 5.11.9.1).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: ConfidenceScore
  parameter_count:
    name: parameter_count
    description: Approximate count of trainable model parameters.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: integer
    minimum_value: 0
  training_duration:
    name: training_duration
    description: Wall-clock duration of model training, expressed as an ISO 8601 duration.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: DurationType
  inference_latency_ms:
    name: inference_latency_ms
    description: Typical end-to-end inference latency in milliseconds.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: float
    minimum_value: 0.0
  model_compression_applied:
    name: model_compression_applied
    description: Whether the model has had compression or distillation applied for
      deployment (Clause 8.6.2).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - AIModel
    range: boolean
  id:
    name: id
    description: Unique CURIE or URI identifying the entity.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    slot_uri: dcterms:identifier
    identifier: true
    owner: AIModel
    domain_of:
    - NamedEntity
    range: uriorcurie
    required: true
  name:
    name: name
    description: Human-readable label for the entity.
    from_schema: https://w3id.org/lmodel/iso22989
    exact_mappings:
    - dcterms:title
    - schema:name
    rank: 1000
    slot_uri: schema:name
    owner: AIModel
    domain_of:
    - NamedEntity
    range: string
    required: true
  description:
    name: description
    description: Free-text description of the entity (paraphrased; verbatim ISO text
      excluded).
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    slot_uri: dcterms:description
    owner: AIModel
    domain_of:
    - NamedEntity
    range: string
  clause_reference:
    name: clause_reference
    description: ISO/IEC 22989:2022 clause identifier (e.g. "5.11.4") the element
      corresponds to.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    owner: AIModel
    domain_of:
    - NamedEntity
    - AutonomyAssessment
    - VerificationValidationFramework
    range: string
    pattern: ^[0-9]+(\.[0-9]+){0,3}$
  aliases:
    name: aliases
    description: Alternative names or synonyms for the term.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    slot_uri: skos:altLabel
    owner: AIModel
    domain_of:
    - NamedEntity
    range: string
    multivalued: true
  preferred_label:
    name: preferred_label
    description: Preferred natural-language label.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    slot_uri: skos:prefLabel
    owner: AIModel
    domain_of:
    - NamedEntity
    range: string
  see_also_uri:
    name: see_also_uri
    description: Pointers to related external resources or term records.
    from_schema: https://w3id.org/lmodel/iso22989
    rank: 1000
    slot_uri: rdfs:seeAlso
    owner: AIModel
    domain_of:
    - NamedEntity
    range: uriorcurie
    multivalued: true