KOMET Unleashed - Ontology

Figure 1: Parts of the ontology of the KOMET Architecture (container classes)

The fact that the overall funcionality of the software system is generated as the sum of all functions of the collaborating problem solving components (solvers), may result in a scenario where solvers must rely on results of other solvers to be able to complete their calculations. If this is the case, solvers must be able to query the information provided by other solvers, check these results for feasibility and use them to compute their calculations.

The KOMET Architecture provides mechanisms for computerised selection and interpretation of information provided by other solvers. Knowledge about the internal data storage concepts of other solvers is not necesary. This is achieved by storage and maintainance of data about data (meta data). Two concepts to store meta data storage are used simultaneously. On one hand, meta data is maintained with help of a relational data base. Which solver provides which information and where this information can be found in the solver data base is stored there at a granularity of tables and columns. On the other hand, a semantic network is built based upon an ontology coded in the Web Ontology Language (OWL). This semantic network describes a taxonomy of forest measurement values and defines among others which information is available within the spatial decision support system, at all. One class within the semantic network is defined for each measurement value. Besides inheritance, further relationships between classes as well as rules can be defined to enhance the description of the structure of the taxonomy. Additionally to the value classes, container classes are defined to realise the posiblility to build logic groups of value classes. These container classes are shown in figure 1.

The solvers define objects based upon the available classes. Thereby, the amount of objects is dependent of type and number of solvers, which are registered in the system. An object oriented data model, which can grow and shrink dynamically is realised this way.

The DSS kernel maintains the meta data by providing several services for these purposes. During registration or un-registration of solvers, value classes are created or deleted respectively. With help of the ontology query service, objects can be selected from the semantic network. Even rules can be incorporated in such ontology queries.

• Relational data storage
• Hig performance and flexibility due to the use of a replaceable RDBMS
• Easy realisation of data access with help of standards (e. g. JDBC and SQL)
• OWL-DL
• flexible means of query definition
• Posibility to realise extremely complex queries by definition of a powerfull system of rules

Martin Döllerer - zuletzt geändert am 01.01.1970 um 01:00 Uhr