- From: Gary Ng <Gary.Ng@networkinference.com>
- Date: Fri, 23 May 2003 13:33:53 +0100
- To: <www-rdf-logic@w3.org>
Georgios, not sure if anyone has replied to this yet. Here is my opinion from a DAML+OIL/OWL point of view. Hope this will give you some insight. Looking at what you have described, not requiring the introduction of instances is important, am I correct? In that case it sounds like what you want is to be able to say: The set of colors that could fill the attribute of Eye - is exactly the same set as - The set of colors that is determined_by Eye_Color_Assay You can say this in DAML+OIL or OWL both of which are based on Description Logic. You need two things: 1. The ability to use the inverse of hasAttribute - isAttributeOf 2. the ability to describe equivalence between classes. 2. is basically an axiom that constrains your ontology, over and above the hierarchy and relationships that you have. Thus in DL: Color AND (exists is_Attribute_Of Eye) -equivalentTo- Color AND (exists determined_by Eye_Color_Assay) ditto for Skin and color. The above has a direct mapping to the serialization syntax of DAML+OIL and OWL. In these languages, since Eye and Skin are all classes, and DL has an open world semantics (i.e. every class COULD be overlapping each other unless you say not so), you also need to say that an Eye is not a Skin (use disjointWith), and Eye_Color_Assay is not Skin_Color_Assay for correct inferencing. Here are some pointers for you: More OWL information http://www.w3.org/2001/sw/WebOnt/ More DL information http://www.dl.kr.org Let me know if you are interested in pursuing OWL further. We are currently working on a visio based tool called Construct which allows you to model this graphically in OWL. It abstracts these and other common axioms into a set of OWL stencils. The output OWL file could also be directly processed using our inference engine Cerebra, which then provides the semantics of the ontology for applications. Cheers, Gary Gary Ng, PhD gary.ng@networkinference.com Senior Software Engineer Network Inference (Holdings) Ltd Tel: +44 (0) 20 7616 0717 Fax: +44 (0) 20 7616 0701
Received on Friday, 23 May 2003 08:35:38 UTC