Subject/predicate/object
From eg
The subject/predicate/object living ontology pattern is best known from RDF, the W3C's Resource Description Framework. Applying this pattern lets you accurately label information so people and search engines can find it. One reason to add semantic tags, simple hidden annotations, is to categorize/spread/uprank- also called SEO. However there are many other good reasons.
To dynamically categorize using semantic tags has some advantages over static categories - though LO clearly specifies a LO category scheme to aid integrativity over the very long term.
Contents |
[edit] explicit tag
Most explicit semantic tags follow the RDF metadata model which makes statements about resources as triples: subject/predicate/objects.
For instance, the content "the bird has the color blue" organized as an RDF triple: Subject = "the bird" Predicate = "has the color" Object = "blue"
The subject is the resource described, the predicate denotes traits or aspects of the subject and expresses a relationship between the subject and the object. Note that a clause using the preposition has must be used to describe an attribute, rather than to be, i.e. in English "the bird is blue" is acceptable, but in RDF and in trollish it is not. See E Prime for why.
OWL/RDF requires the use of URIs as XML ids. However there are a number of ways to specify SPO without using XML directly at all.
[edit] as used in semantic mediawiki
A convention for semantic wiki links has been established de facto by semantic mediawiki, well described by wikicompany and the use of which for SEO via Centiare. SMW simplifies use of RDF as "some OWL individual/class/role are specified in Semantic MediaWiki" [1].
The many projects using explicit semantic tags in mediawiki support a wide variety of microformats (called semantic templates) and related conventions like:
- timelines
- service links, e.g. an "article Attribute:Coordinates might include the annotation online maps" [2]
Some explicitly political public wikis use it:
and some startup businesses:
- wikicompany (business listings)
- Centiare essentially a for-profit owned-page listing
A few seem to be seeking consensus across multiple languages:
- Auto terms in English, German, Turkish
- jurispedia (Arabic) and (French) with US and UK coming - all of which provide RDF output as CC-by-sa!
Some focus on software development and code and may achieve some standards for case roles and other UML constructs:
Some projects like DODA are explicitly devoted to lightweight integration of semantic data and use use case analysis internally, so are likely to set standards in this regard.
Development of specific tags is not a priority for semantic mediawiki. The development of semantic mediawiki itself focuses on the facility, rather than for instance freebase .com or OntoWiki which make instantiation of fixed schemas easy. Semantic mediawiki is by contrast "about unstructured data, that can be enhanced with some structure between blobs of unstructured data, basically, text." All seem to be converging on subsets of SPARQL, the de facto standard query language for RDF.
[edit] LO implicit tag
Another approach to lightweight integration is to avoid using RDF until the last possible moment, and avoid even using mediawiki-specific tags.
The Living Ontology Web relies on many implicit semantic tags, e.g. a link to a page with the name by Craig Hubley implies the predicate by and the object Craig Hubley, with the preceding name as the anchor or subject. These may be translated into hidden explicit tags, but only need to be if the semantics varies from the default meaning.
An implicitly or somewhat semantic wiki has many such implicit tags. A LOWest Troll is required to maintain them and discipline users who damage them.
[edit] LO audit
An LO audit establishes the false inference impact of a given semantic wiki's tags, a calculation which is not easy to make. See LO-compliant. Note: the W3C definition of the semantic web assumes a false inference impact of infinity, while the actual WWW assumes a false inference impact of zero. For now, LO-compliant is defined as assuming false inference impact of 1, nominally set as equivalent to a civil lawsuit loss of one dollar.
[edit] documented as threefolds and compiled into triples
Many kinds of subjects, predicates, objects are described in living ontology, typically as living ontology patterns. The pattern itself does not include all the information, usually, to translate the information directly to explicit tags - for instance faction/role/order is not in this form, and while the faction is the subject, the order is issued by the faction and is accordingly the object, the role is not a predicate - rather, the role defines a range of possible predicates.
That is, the pattern is not a triple in the RDF sense but rather a threefold that implies many predicates, those that arise from a role which takes the orders. As this example implies, deference relations must be resolved before any references, and these must be resolved in order to trust in any inferences. The calculated false inference impact is from the perspective of the user and taks themselves, the references are resolved from page names into semantic tags, and only after that RDF style processing occurs. This implies that each user is an authority, each page is a source, and semantic tags characterize evidence so a
TIPAESA structure can be created to characterize the motives for the action.
[edit] Sources
- W3C RDF Primer A basic RDF technical primer describing:
- Effective usage and basic concepts such as XML syntax;
- How a Vocabulary Description Language defines vocabularies;
- An overview of some deployed applications;
- Content and purpose of related specifications.
- Basic annotation features of semantic mediawiki
