I remember that Elmo is very nicely done, using only interfaces for class definitions, and java-proxy objects for the implementations allow quite clean multiple inheritance...
In general though I wonder if these frameworks really are worth it. The theory is that they hide the "complexity" of RDF and allow normal java programmers to work with RDF - but in the process they also hide one of the main benefits of RDF, being able to deal with data-structured not known at compile time.
The "schemalessness" of RDF is great for small projects, or for extending normal schemas, and of course RDFS offers a way to formalise and standardise your new properties/classes when you need to. As Kendall Clark says "Correctness when it matters; sloppiness when it doesn’t". If you pre-generate all your object interfaces from a schema you lose all this - of course you could have a generic bean interface at the top (i.e. DynaBean) which would let you query for properties not in your scheam, but then you could just as well just use the Jena OntResource.
发布评论
评论（1）
还有RDF2Go，它有RDFReactor< /a>.
我记得 Elmo 做得非常好，仅使用接口进行类定义，并且 java 代理对象用于实现允许相当干净的多重继承......
总的来说，尽管我想知道这些框架是否真的值得。理论上，它们隐藏了 RDF 的“复杂性”并允许普通 java 程序员使用 RDF - 但在此过程中它们也隐藏了 RDF 的主要优点之一，能够处理编译时未知的数据结构。
RDF 的“无模式”非常适合小型项目或扩展普通模式，当然 RDFS 提供了一种在您需要时形式化和标准化新属性/类的方法。正如肯德尔·克拉克所说“正确性不重要时的草率”。如果您从模式中预先生成所有对象接口，您将失去所有这些 - 当然您可以在顶部有一个通用 bean 接口（即 DynaBean），它可以让您查询不在您的架构中的属性，但您也可以只使用 Jena OntResource。
There is also RDF2Go, which has has RDFReactor.
I remember that Elmo is very nicely done, using only interfaces for class definitions, and java-proxy objects for the implementations allow quite clean multiple inheritance...
In general though I wonder if these frameworks really are worth it. The theory is that they hide the "complexity" of RDF and allow normal java programmers to work with RDF - but in the process they also hide one of the main benefits of RDF, being able to deal with data-structured not known at compile time.
The "schemalessness" of RDF is great for small projects, or for extending normal schemas, and of course RDFS offers a way to formalise and standardise your new properties/classes when you need to. As Kendall Clark says "Correctness when it matters; sloppiness when it doesn’t". If you pre-generate all your object interfaces from a schema you lose all this - of course you could have a generic bean interface at the top (i.e. DynaBean) which would let you query for properties not in your scheam, but then you could just as well just use the Jena OntResource.