A DSL for describing and deducing relationships between concept terms.
This project is based on the Chevrotain parser building toolkit, and meta-chevrotain which is a DSL for generating chevrotain parser
This DSL can be used to describe a context of related concept terms, usually names of sets, categories, or tags.
Then, it can deduce whether a object in such context is valid, and if it's valid, its implied relations of all concepts in the context.
- Map Item
- Anything on a map is a Map Item
- Structural and Geographic
- A Map Item is either
- Structural, meaning it's a concept related to some structure: house, road, sign, etc., or
- Geographic, meaning it's a geographical concept: island, lake, continent, etc.
- A Map Item is either
- Building
- All Buildings are Structural
- House
- All Houses are Building
- Transportation
- All Transportation are Structural
- Road
- All Roads are Transportation
- Lake, River, and Ocean
- All Lakes, Rivers, and Ocean are Geographic
- Water
- All Lakes, Rivers, and Ocean are Water
The above relations can be described as the following code
?@Mode
>- map_item {
~^ geographic, structural
}
>+ structural {
>+ building {
house
}
>+ transportation {
road
}
}
>+ geographic {
>+ water {
lake
river
ocean
}
}
The syntax has a directed-graph-like structure, where the nodes are the concept terms, and their position in the structure denotes their relation.
The direction of the graph edge is determined by the ">" and "<" order prefixes. For example,
>+ building {
house
}
defines an edge from Building to House. If the ">" symbol is replaced with "<", the edge direction would be reversed.
The nodes that has a outgoing edge are called Declarations, which means they declare certain relations for the nodes they points to.
The nodes that has no outgoing edge are called Qualifications, which means they defines some conditions, such that once a object qualifies those conditions, the object are thus subject to the bounds of all declarations declared upon the qualification.
The two key relations between concepts terms are "is" and "isnt", which are denoted with the "+" and "-" verb prefixes in declarations.
Back to the previous example, in
>+ building {
house
}
there is one declaration and one qualification.
The declaration ">+ building" declares whatever it points to must be "is building".
The qualification "+ house" qualifies whatever object that satisfies "is house".
Therefore, this example as a whole means "All houses are building"
Unlike declarations which can only have one term, qualifications can have a list of term along with a list prefix denoting the how the list is interpreted.
There are three logical prefixes, "&" for And, | for Or, and ^ for Oneof. For example, if one wants to say "anything that is both Building and Deserted are ToBeImproved", the following can work
>+ to_be_improved {
&+ building, deserted
}
There is also another prefix for qualifications, the negation prefix "~", which reverse-selects what the qualification originally matches. For example,
>- map_item {
~^ geographic, structural
}
says "anything that is not one of geographic and structural is not a map item"
Just noticed the existance of RDF, RDFS, and OWL, which fulfills the intention of this project.
Development is suspended, and project is archived.