Authors: Vadim Ermolayev
Abstract
The idea of the presented approach is to borrow a plausible analogy of a "system law" As remarked in [2], a system law is a rule which generalizes the behavior of some observed phenomenon within a concrete system and its given spatiotemporal context. A system law tells what behavior is expected within the system. Thus a system's law can cause change or represent a barrier to change. It can be used to predict certain aspects of the system behavior, which are based on the force, or influence it exerts on the internal environment of the system. In contrast to a natural law, a system law is neither universal nor does it need to be true, correct, etc. from the field of Dynamics in Mechanics - the Newton's Law of Universal Gravitation. This analogy is exploited for building the law of gravitation in dynamic systems comprising a Domain of Discourse and knowledge representations (ontologies) describing this domain. As ontology elements do not possess physical mass, this component of the gravitation law is substituted by the property of fitness of an ontology to the requirements of the knowledge stakeholders characteristic for the described domain. It is also argued that the implementation of the developed theoretical framework is feasible as the supporting techniques, including some software tools, already exist. As the examples of the relevant component methods and tools, the paper presents concisely the OntoElect methodology, Ontology Difference Visualizer, and Structural Difference Discovery Engine. These instruments help solve some practical problems in eliciting domain requirements, developing structural contexts for the requirements, generating the mappings between these structural contexts and the target ontology, computing increments and decrements of ontology fitness based on these mappings. It is concluded that the presented framework has prospects to be applied practically for visualization and analysis of ontology changes in dynamics. Use cases for ontology refinement and anomaly detection are suggested for validation.
Fulltext
–
0.52 Mb