Change - SemaOntology Desic Gapgn

Created on March 13, 2013, 8:21 a.m. by Hevok & updated on March 13, 2013, 8:48 a.m. by Hevok

In general the Interpoerability of Metadata should be achieved by the use of Ontologies. ¶
The Problem is among heterogenous Metadata there is a Semantic Gap. This Semantic Gap can be bridged with the Help of Ontologies. The Problem of the Semantic Gap means the different Ontologies can be applied to represent identical Knowledge. The Problem is how can one find out that these different Ontologies really mean the same Thing or describe the same Thing. ¶

Even cosidering a small world with three bricks in an initial state and one wnats to bring this bricks into a final state in the same world, where all of these bricks stand one of Top of each other. ¶

One can model this with some Objects (Block A, B and C as well as a Table and a hand that is able to grab these Objects). Then one has several Relations, for example there is the on Relation that says a block stands on top of another block. There is a clear Relation that means if we have a block in the hend we clear it. The block is hold with the holding Relation. Lastly one cna ask whether the hand is empty or not. With this modeling one is able to bring the world from its initial state to the final state of the world. ¶

Tis is however not the only way to model this. Another approach is with less objects, namley one Block A, B and C. So here we do notneed the table nor the hand, but therefore we have other kind of relations. In addition to the on, clear, hlding and handEmpty Relation there is instead of defining a table a Relation that a Block X stands on a table (which is other true or not true). ¶

To model this in kind of a represential Tree one has Entity and Relation Classes. There one has Entity Table, block and hand. Beloning to them there is a table T, Blocks A,B and C as well as hand H. The Relations can be binary, unary or even have no arguments like handEmpy which gives back a boolean (True or False). The World can also be described with the help of Axioms where one says if X stands on Y and Y stands on Z than X is above Z for example. ¶

In the other model one has only one type of Entity, namely block which contains Block A, B and C. There are also Relations with an additional unary Relation onTable. ¶

The problem is that one has two different models, but from the tree represention they are quite similiar. We can identify in these trees which has the same Meaning or does represent the same as in the other tree. So one can define Mappings. One has to do this because one wants to find out whether this two Models really denote the same. To brigth this we need an Ontology representation that is able to map one model into another Model. One has to understand this Model. ¶

The Semantic Web is based on interoerability of Metadata
Among heterogenous Metadata there is a Semantic Gap that can be bridged with the help of Ontologies

- Proble of the Semantic Gap: ¶
+ differnt Ontologies can be applied to represent identical Knowledge. ¶

Let's Model a world: ¶

- Models: ¶

Objects Relations Objects Relations
block A on(X,Y) block A on(X,Y) ¶
block B clear(X,Y) block B clear(X) ¶
block C holding(X) block C onTable(X) ¶
table T handEmpty holding(X) ¶
hand H handEmpty ¶

- Axoim: on(X,Y) ⋀ on(Y,Z) → above(X,Y) ¶

behind the Model there is an Ontology
Ontology Developement is an interative Process, which means step by step one refines and extends the Ontology. Of course, sometimes one has to go back. So it is an iterative Process that repeats continously and improves the Ontology by doing this. ¶

There is not an only single way to develop an Ontology, in fact there are different ways for modeling an Ontology. In practice it is always the designated Application that decides about which modelling Approach to choose, i.e. which cycle to repeat and how to go back in this development Process. ¶

There must also be place for the Formulation of complex Axioms including Property Constrains but one also needs to make Rules or other type of Constrains concerning the Classes that they should obey, then one should consider to include the definition of Axioms that are needed for the Knowledge Base and the Ontology and Knowledge one wants to represent. ¶

On the other hand the whole process has to fit for large scale industrial Ontology Engineering, including version Control strict Evalution and Quality Assurance in a professional Engineering Process. ¶

Thus Ontology Development is an interative Proess where one goes back if one has forgotten to refine soemthing and one continously to improve the Ontology. ¶

There is not one correct way to model a Domain and there are always viable alternatives. ¶

in practice an iterative process that repeats continuosly and improves the Ontology
there are always different Approaches for modeling an Ontology
in practice the designated Application decides about the modeling Approach ¶

Attention: ¶
- Futher refinement should include the Formulation of Axioms
- Should be well suited for large scale industrial Ontology Engineering
+ Version Control, Evaluation, Quality Assurance, etc. ¶

"There is no one correct way to model a domain. There are always viable Alternatives."

Parent: Semantics Ontology

Comment: Updated entry

Comment on This Data Unit