如何在RDF词汇表中定义自定义受限数据类型？有无标准方案？

Great question—this is a common point of confusion because RDF/OWL gives you a few options, but not all are equally standard or tool-friendly. Let’s break down the recommended practices:

1. OWL Datatype Restrictions (Official Standard) #

The Turtle snippet you shared using owl:onDatatype and owl:withRestrictions is actually the W3C-recommended way to define constrained datatypes in OWL. This approach is explicitly part of the OWL 2 specification, and it’s recognized by RDF reasoners, validators, and tools.

Here’s a polished example of your equipment category name constraint, formatted properly:

:EquipmentCategoryName a owl:DatatypeProperty ;
    rdfs:domain :EquipmentCategory ;
    rdfs:range [
        a rdfs:Datatype ;
        owl:onDatatype xsd:string ;
        owl:withRestrictions ( 
            [ xsd:maxLength "10"^^xsd:int ] 
            [ xsd:minLength "1"^^xsd:int ]
        )
    ] .

You can extend this with other XSD facets too—like xsd:pattern for regex validation, xsd:minInclusive for numeric ranges, etc. The anonymous datatype here is treated as a proper restricted subtype of xsd:string by compliant tools.

2. XSD Custom Simple Types (Interoperable with XML) #

Your XML Schema example is another valid approach, especially if you need interoperability with XML systems. To use this in RDF, you just need to:

• Define your custom XSD type (like your Char_20 type) in a schema file
• Import that schema into your RDF vocabulary’s namespace
• Reference the custom type directly as a property’s range:

  :shortName a owl:DatatypeProperty ;
      rdfs:range my:Char_20 .
  

This works because RDF natively supports referencing XSD datatypes, including custom ones. It’s a good choice if you’re working in an environment where XML and RDF coexist.

3. Why rdfs:subClassOf xsd:string Is Not Standard #

You’re right to be skeptical about this approach. In RDF/OWL, rdfs:subClassOf is meant for classes (instances of owl:Class), while xsd:string is a datatype (instance of rdfs:Datatype). These are distinct categories in the RDFS/OWL semantics.

While you can write the syntax, reasoners and tools won’t interpret this as a datatype constraint—they’ll treat it as an incorrect cross-category subclass relationship, which can lead to unexpected behavior or ignored constraints. So this is definitely not a recommended practice.

What About schema.org? #

schema.org’s design prioritizes simplicity and broad adoption over complex constraints. By sticking to basic XSD types, they lower the barrier for non-expert users and ensure compatibility with search engines and lightweight tools. This doesn’t mean custom restricted datatypes lack standards—it just means schema.org chose a simplified path for its specific use case.

Final Recommendations #

• For OWL vocabularies: Use the owl:onDatatype/owl:withRestrictions pattern. It’s the most tool-friendly and semantically correct standard.
• For XML-interoperable vocabularies: Define custom XSD simple types and reference them in your RDF.
• Avoid: Using rdfs:subClassOf with datatypes—it’s semantically invalid and unsupported by most tools.

内容的提问来源于stack exchange，提问作者inf3rno