Semantic tags

Attach icon tags to text nodes in order to define semantics.

The image above generates the following html code: personLikesAnimalOntology Living Person likes Animal Ann likes Hoppel Animal Schlappi Hoppel

Note that we get only one inline par because we didn't add any text structuring information. If we strive at a more layouted representation of our little taxonomy we can bring together semantic tags and edge taggings, like e.g.

This results in the following html representation:

• personLikesAnimalOntology2
  • Living
    • Person likes Animal
      • Ann likes Hoppel
    • Animal
      • Hoppel

Additionally there is "real" semantics generated: (This code is shown if you follow the personLikesAnimalOntology link in the online version of this document.)

                           
Person rdfs:subClassOf  Living . 
Person rdfs:domain likes . 
likes rdfs:range Animal . 
Ann rdf:type Person . 
Ann likes Hoppel . 
Animal rdfs:subClassOf Living . 
Hoppel rdf:type Animal . 
                        

In order to support readability of the generated code for a broad publicum we focus on exporting semantics in turtle syntax. However, there is a commercial version of semAuth available which also exports OWL2 XML and F-logic files. We have reserved this feature for trained ontology engineering professionals because OWL modelling much more complex then RDFS modelling.

Targeting at end users there is an XML export included in the semAuth RDF version which supports semantic authors to review and communicate their semantic models: semAuth also generates a GraphML representation of all valid ontology objects encountered so far. For semAuth is not a graph layout tool our GraphML representation completely lacks layout information. However, this output can be loaded i.e. into the free version of the yEd graph editor (http://www.yworks.com/en/products_yed_about.html). yEd provides a broad range of highly sophisticated graph outlining features. After some supervised auto layout steps we may yield e.g.:

Modularizing the GraphML output from the very layout process allows the end user to play with various visualizations. While we already suspected that every ontology design pattern benefits from it's own specific visualization handling playing with yEd provided a new experience to us: Often it is the play of finding an adequate ontology visualization which provides more insight into a knowledge domain than any other analysis. (And yes: A GraphML to semAuth filter which will allow for a roundtrip between a graph and an mindmap view is already planned.)

We included the GraphML export into the GPL version of semAuth because the possibilty of playing with many very different views ("media") from one single source perhaps is the most important feature of semAuth.

Summary: Semantic tags are icons which are attached to nodes. They declare a node being a class, an instance, a data or an object property etc. RDF riples are generated by interpreting chains of semantically tagged icons. This is the full set of supported icon chains and the generated turtle triples: ../xslt/describeConfig.html

Implementation detail: As shown in the example personLikesAnimalOntology above you can substitute every iconic tag with textual tags like (i), (C), (->) etc. With this feature you can author semantically tagged trees in your favourite text editor. If receiving text from copy and paste freemind interprets indentations appropriately to reproduce a tree.