AML (Anything Modeling Language) is a declarative language for defining metadata documents that can be parsed into graphs of information. These documents can be published and linked on the web or through any other type of API.
It can be used to describe common, existing YAML/JSON metadata documents or to define the semantics and structure of completely new metadata formats.
AML metadata documents have the following advantages:
- Well-defined, formal semantics that can be reused across multiple types of metadata
- Explicit schema that can be used to automate parsing and validation
- Support for hyperlinks as a mechanism to connect and publish metadata documents
- Out-of-the-box modularity, allowing the metadata graph to be split into composable sets of documents
- Direct translation into a graph of data that can be stored, queried and published
- Support for W3C recommendations for the web and metadata
Defining metadata documents
To define a new type of metadata document, an AML dialect must be used to map the structure of a YAML/JSON document to a graph of information and overlay the semantics for the nodes in the graph. With that information, an AML processor tool can parse and validate the instances of the metadata document and generate the output graph of information with the defined semantics.
This is a simple example of using the AML dialect to define a personal information metadata document:
dialect: Person Card
With this very simple metadata document, an AML processor can validate and parse metadata descriptions according to the defined "Person Card" format:
#%Person Card 1.0
name: Chuck Berry
The outcome of parsing this example is a JSON-LD document, encoding a graph with the information about the people described in the "Person Card" documents:
"schema-org:name": "Chuck Berry"
This graph of information can be saved in JSON-LD compatible graph databases, or it can be queried using a standard graph query language.
In this example, the semantics of the information comes from the Schema.org project. Schema.org provides detailed definitions like Person that can be processed by machines and re-used across multiple metadata documents even if the syntactical details to describe a person varies.
AML dialects describing metadata documents can work with any of these existing vocabularies as long as the concepts in the vocabularies can be identified by a URI.
AML also introduces the AML vocabulary specification to describe new semantics that can be re-used across AML dialects.
For example, we could introduce our own "Personal Information" AML vocabulary with our own definitions for classes of entities and properties:
vocabulary: Personal Information Management
description: Person being described
description: date of birth of a person
description: full name including surnames
AML vocabularies can be used in AML dialect definitions and mixed with external vocabularies.