more (structured) data

[nsajko]

Currently this package attempts to associate each well-known and important graph property or class of graphs with a very short description in natural language.

It might make sense to provide more machine-readable data. Associate each well-known and important graph property or class of graphs with:

• A machine-readable name (identifier in a programming language).

• A set of alternative names.

• Assumed properties, by which I mean the other properties that are required to be satisfied before considering this property is valid. For example, the number of connected components is not defined for directed graphs (as opposed to, for example, the number of strongly connected components), so the property should be stored together with a (machine-readable) note saying that the property is only meaningful for undirected graphs.

Then the "very short description in natural language" would perhaps not be necessary any more?

It would not (?) be necessary to expose this extra data externally, however giving the data more structure would make it easier to check everything is correct (the doc strings here and the dependent test suites) and to add more entries.

xref:

• interoperability with ISGCI, HOPS, FindStat, Wikidata #18

• store the complete data as JSON, to make it useful for other programming languages, too? #19