The Description of a Model (DOAM) risk model ontology, described using W3C RDF Schema and the Web Ontology Language OWL.
DOAM is used in the Open Risk Manual to organize the semantic annotation of risk model concepts
DOAM was inspired by the Description of a Project (DOAP) vocubulary developed by Edd Dumbill that was used for the annotation of open source python applications
Open Risk
The Description of a Model (DOAM) risk model ontology, described using W3C RDF Schema and the Web Ontology Language OWL.
application/rdf+xml
Copyright © 2015-2021 Open Risk
Description of a Model (DOAM). A Risk Model Ontology
We welcome contributors to the Risk Model Ontology!
Creative Commons Attribution 3.0 (CC BY 3.0)
v0.4 Introduces PMML semantics
v0.3.2 Split out Risk Data related concepts into the RDO ontology
v0.3 Significant expansion of the Risk Model Ontology to capture different model aspects in more refined manner. Version 0.3 captures the distinction of Abstract Models, Specified Models, Model Source Code and Model Instances
v0.2.8 First reasonably complete release
Draft
The Risk Model Ontology is still in active development. Feedback, ideas, use cases are encouraged
0.4
has abstract model
1
Each risk Model must have exactly one Abstract Model Representation
1
Any person that is the author of an abstract model specification (e.g. a paper where the model is specified, conceptually and mathematically)
has author
has model definition
Developer is any person who contributed to the development of a full specification and implementation of an abstract model. Has authored segments of its source code distribution
has developer
Contributor to the documentation of the model implementation
has documenter
Maintainer of a source code distribution of given model
has maintainer
The source code distribution that implements the abstract model
has model implementation
has model instance
has model specification
The source code repository of a model implementation
has repository
The scope of the model application (type of product, portfolio etc.)
has application scope
A tester is a programmer or other qualified quality control person that is responsible for the
integrity of the source code distribution of a model implementation or a model instance
has implementation tester
An analyst or other qualified individual providing peer review and validation of an abstract model specification
has model validator
An abstract model description that the source code implements. It is a reference to a fully
defined abstract linked model
implements model
The model category (classification by risk model type)
belongs to risk category
The date on which a particular abstract model or source code distribution is released
A Model Instance does not have a creation date as instances will have a regular down/up cycle. Usage of model instances tracked via user workflows
For Abstract Models, the creation date indicates when the model has becomed linked with the formal documentation available in a linked URI. It does not refer to the time of first publication.
creation date
data end point
Description of the model, code distribution or instance
description
The location where documentation about the model or its implementation can be found
documentation URL
The URI of a web resource where input data required for a model calculation are stored
input data end point
input data string
The name of the risk model maybe different from the underlying mathematical model
The name of a model, release etc.
risk model name
The URI of a web service endpoint where the data outputs of a Model Instance can be stored
output data end point
output data string
publication date
repository URL
The revision number of a model or source code version release
revision
The URI of a web service endpoint where the Model Instance implementing an Abstract Model can be accessed
service end point
Short description of the model
short description
The URL at which online training resources regarding an abstract model, its source code or a model instance can be found
training resource URL
foaf:Person
The theoretical (abstract) description of a distinct risk model. It is expressed as a document (paper, wiki page) that holds the human readable model description (possibly including mathematical notation)
The abstract model name can best be captured as the name of a vector valued function (with multiple possible inputs and outputs)
Abstract Model
Anomaly Detection Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Anomaly_Detection_Model
0.4.1
Association Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Association_Model
0.4.1
Baseline Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Baseline_Model
0.4.1
Bayesian Network Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Bayesian_Network_Model
0.4.1
Clustering Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Clustering_Model
0.4.1
Data Dictionary
Decision Tree is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Decision_Tree
0.4.1
Gaussian Process Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Gaussian_Process_Model
0.4.1
General Regression Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/General_Regression_Model
0.4.1
Git source code branch
Git Branch
Mining Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Mining_Model
0.4.1
Mining Build Task
The source code distribution and any other artefacts implementing an abstract model
The name of the model source code is e.g. the name of the directory that contains the distribution. May include version information
Model Implementation
A web application instance that makes an abstract model available by executing its programmatic implementation (whether compiled or interpreted)
While risk models can be implemented in various forms, only risk models that are accessible (in-principle) online via a URL are considered - in line with the Risk Model Ontology being a web OWL ontology
Model Instance
The full specification of an abstract risk model that
1) identifies and links model inputs to concrete and observable risk data sources.
2) expresses the model structure via e.g., enumerations of variable names, functional forms etc
The abstract model name can best be captured as the name of a vector valued function (with multiple possible inputs and outputs)
Model Specification
Model Component
Naive Bayes Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Naive_Bayes_Model
0.4.1
Nearest Neighbor Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Nearest_Neighbor_Model
0.4.1
Neural Network is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Neural_Network
0.4.1
The specification of an Abstract Model using the PMML (Predictive Model Markup Language)
PMML Specification
A predictive model (e.g., specified via the PMML standard)
Predictive Model
Regression is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Regression
0.4.1
Regression Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Regression_Model
0.4.1
Source code repository for the model
NB: Might migrate the concept to external description of software repositories as this is not unique to the risk management domain
Repository
The root class capturing all the elements of a risk model (theoretical concept, implementation and deployed instances)
Risk Model
Risk category to which this model belongs (type of risk that is being modeled)
Placeholder. For forward compatibility to link with risk taxonomy ontologies
Risk Category
Rule Set Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Rule_Set_Model
0.4.1
Scope of model (application domain). The type of entities (within a broader domain) that are covered by the model
Placeholder. For forward compatibility with scope taxonomies
Model Scope
Scorecard is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Scorecard
0.4.1
Sequence Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Sequence_Model
0.4.1
Support Vector Machine Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Support_Vector_Machine_Model
0.4.1
Text Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Text_Model
0.4.1
Time Series Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Time_Series_Model
0.4.1
Transformation Dictionary
Tree Model is one of the defined predictive models of the PMML standard (4.4)
https://www.openriskmanual.org/wiki/Tree_Model
0.4.1
Version information of a published abstract model or source code release
Might migrate as not unique to this use case
Model Version