The emerging Semantic Web will revolutionize current approaches to modeling and simulation. Under this model all of the data needed for simulations will be stored on public and/or private webs. This data will be semantically represented as ontologies using the W3C Web Ontology Language (OWL). The following paragraphs provide an introduction to the Semantic Web, OWL, ontologies and the concept for how these will impact modeling and simulation in the future.
The Semantic Web is an initiative sponsored by the World Wide Web Consortium (W3C). The Semantic Web is an extension of the current Web where information on web pages is semantically represented. The most meaningful way to represent the meaning of of information is through the use of relationships. These relationships are expressed as ontologies that provide a formal, machine readable, representation of the meaning of the concepts expressed on web pages. The standard language for writing web-based ontologies is OWL, which is an official W3C recommendation. While the Semantic Web will be used to represent all sorts of knowledge, one of its powerful benefits will be to represent knowledge relevant to computer simulations. Currently, simulations rely on information/knowledge that is stored in files, tables, data bases or hard-wired into the simulation code itself. The concept of Semantic Simulations is to represent all of the relevant simulation information/knowledge on the Semantic Web. This will allow simulation code to access the specific information needed at any point in the simulation directly from the Web. Additionally, the fact that the ontologies that will contain the relevant knowledge are directly machine readable and understandable, will greatly reduce the level of assumptions that are needed in the simulation execution software. Traditional simulation software will evolve into simulation agent software. The simulation agents will execute using the simulation data accessible via the Semantic Web. Simulation agents will control what is to be done with the semantic data. The Semantic Web will provide a natural separation between the software to execute data and the data itself. In that way the semantic data can be reused without modification by many different set of execution code that will accomplish different objectives. This will allow for huge amounts of source data, spread across literally millions of semantic web servers, that will be accessible by simulation developers to use in a wide variety of ways. It will also allow for the easy addition of new semantic data to further enrich existing simulations.
Ontologies are much easier for most people to understand when they are represented in a graphical format. To facilitate the graphical representation of ontologies written in DAML, the immediate precursor of OWL, I created VisioDAML. Follow the VisoDAML link to find out how to use this graphical tool and to see examples of graphical ontology representations. I have also developed VisioOWL, which supplants VisioDAML. A simple example of a graphical visualization of an ontology is shown below.
Military Vehicle Ontology: MilitaryVehicles have DataTypeProperties that are inherited by their subclasses, and the subclasses have ObjectProperties that they may or may not share with other subclasses.
Note that the OWL class MilitaryVehicle has several data type properties: name, weight, topSpeed and personnelCapacity. The sub-classes: Tank, Airplane, Truck and Ship inherit these properties. The subclass Tank has object type properties: weapons, turret, track and engine. The object property “engine” is also shared by the Airplane, Truck and Ship classes and is actually described by another ontology called Engine that is remotely located at the web site http://site.military/engine-ont. Also note that there is another ontology of a Tank with a subclass of ArmoredCarrier remotely located at http://site.mil/depot-ont. The Tank classes at both sites are defined as the same class. This simple example also illustrates a very powerful feature of ontologies – they support inferencing. In this example the fact that an ArmoredCarrier is a MilitaryVehicle can be inferred from the two ontologies even though there is never an explicit statement that an ArmoredCarrier is a subclass of MilitaryVehicle.
OWL ontologies are designed to live on the Web. They are readable and understandable directly by software programs, and they provide semantic meaning for the concepts they depict. You can provide a web pointer to the associated ontology (http://army.mil/tank-ontology) so that your software application will clearly understand the semantics of the word “tank”. These web pointers to ontologies are called Universal Reference Indicators (URI), a more generalized version of the familiar Uniform Reference Locator (URL) used to point to HTML web pages.
Another reason to express the semantics for concepts as web-based ontologies is to eliminate the need to build the understanding of military vehicles, for example, into every simulation or other application that might need those semantics. By pulling the semantic description up to the Web level where it can be referenced by a URI, many different software applications can be made simpler (lighter weight) by taking advantage of the semantic descriptions available on the Web. Additionally, related applications will consistently use the same semantic description so that they will be interoperable in how they perceive objects.