Advanced Metadata in E-Government
October 19th, 2005
RDF, like XML is an open standard from the World Wide Web Consortium (W3C) that is well-supported with free software. In RDF, statements are encoded as as (subject, predicate, object) triples. The RDF data model is based on the concept of a statement and the concept of `quotation‘ (or reification) of statements. An example of the use of RDF to define metadata is the Dublin Core initiative.
Quotation can be used to make statements about (reified) statements — the statement is treated as the subject or object of another statement. Quotation allows one to create trust, because it allows one to express where something was stated, who stated it, who modeled a statement made by someone, and what level of guarantee they dare to associate with it. `Syndication’ of information is based on this notion of trust. You may be willing to pay for advice, for instance, if you trust the party that gives it even if advice on the same subject is also available for free from other parties.
RDF Schema and OWL
In addition to the RDF data model, additional RDF Schema and Web Ontology Language (OWL) statements can be defined. Both RDF Schema and OWL include the RDF model, but pose restrictions to it which make it useful for defining and describing statements in terms of classes and restrictions on properties, and allow the use of reasoners (e.g. FACT++, Racer, Pellet, …) to perform standard inferencing on models. OWL is typically used to build ontologies (e.g. LKIF-Core), which define the concepts of a certain domain.
So what does this have to do with Legal Texts?
By integrating MetaLex with the semantic web standards RDF, RDFS and OWL, we can transparently describe metadata both on the elements of legal texts themselves, and on the contents of those texts. For example, RDF structures can be used to integrate different manifestations of the same legal text, by using the standard Uniform Resource Identifier (URI) of MetaLex elements. Furthermore, RDF can be used to capture elaborate version management issues. OWL and RDFS can be used to describe the contents of legal texts: the concepts, but also their normative content. These semantic representations of legal texts can be used to performe elaborate legal reasoning, such as consistency checking, legal assessment etc. and for building knowledge-based applications which can be used by citizens to gain advice on complex legal issues.
The advantage of integrating the syntactic structure (i.e. plain MetaLex XML) with the semantic structure is that changes to a text can be immediately propagated to the depending (i.e. referring) texts and applications: an advanced legal content management system (see e.g. this paper).
MetaLex2OWL
The MetaLex distribution contains a stylesheet for translating plain MetaLex to its RDF/OWL representation. This stylesheet resides in the transform directory of the distribution. To see the result of applying this stylesheet, please have a look at the following two versions of the Rome Statute:
- The Statute in plain MetaLex
- The Statute in MetaLex OWL
- No instances should be untyped. If there are untyped instances in the loaded RDF/OWL environment, then the OWL environment is not OWL DL compliant. If your MetaLex file refers to external sources, then give these sources a type. In the case of Cite links, the target of the link is a
metalex:aSource, and in the case of Reference links, it is anowl:Thing. - The document should have a set
xml:baseattribute on therdf:RDFelement. - The
rdf:aboutattribute on the Ontology element should be set to either the exact value ofxml:base, or the empty string (rdf:about=""). Setting an empty string however can cause problems in other XML applications. - The
xml:baseset in each file loaded in the OWL environment should be different for some reason.
Giving each example OWL file generated from this website a unique base is very problematic. To put things in perspective: every resource mentioned in the OWL files that can be generated by this website has a unique, fully qualified URI (i.e. we never use rdf:ID). The xml:base is therefore not at all necessary for identifying resources, but only for qualifying the (unnamed) RDF triples themselves. Demanding an unambiguous identity for triples is obviously silly from a semantics point of view: if you reify a triple by giving them identity, you then have to establish the identity of the triples you use to describe the reified triple, leading to an OWL file of unlimited size.
The OWL generated by this website sets the xml:base and rdf:about to the xml:base of the MetaLex document followed by /owl. It will therefore load into Protege and TopBraid. Files generated with the stylesheet in the distribution zip file will not, but the stylesheet can be easily adapted using the information above. Adapting or extending the stylesheet is recommended anyway: the default stylesheet will for instance also strip your MetaLex file from any MetaData elements containing metadata from other XML schemas.
Entry Filed under: Examples
Trackback this post