Semi-automatic approach to domain ontology building
2013
Gorskis, H., Riga Technical Univ. (Latvia). Faculty of Computer Science and Information Technology. Dept. of Modelling and Simulation | Zmanovska, T., Riga Technical Univ. (Latvia). Faculty of Computer Science and Information Technology. Dept. of Modelling and Simulation | Chizhov, J., Riga Technical Univ. (Latvia). Faculty of Computer Science and Information Technology. Dept. of Modelling and Simulation
This paper presents an automated method for building task ontology models from guideline models. A guideline is a specification of steps that need to be taken in certain situations and criteria that need to be fulfilled for these steps to be chosen. The ontology building process concentrates on machine readable guideline models, in particular, on the Guideline Interchange Format (GLIF). Since guidelines for similar processes most likely have many common concepts, it can be proposed that an ontological model of the task domain could be used for information storage and rule extraction. In order to accomplish the set goal of building a task ontology model from guidelines, it is necessary to do the following: create or convert basic concepts of the task into the concepts of ontology; create a relational structure between concepts within the ontology, capable of representing the choices and the sequence of the guidelines and unify equal concepts from different guidelines. The extraction of concepts can be done by finding data request and task execution blocks in the guideline model. Data request blocks would correspond to environmental or system state concepts. To create a class hierarchy of the tasks and concepts for the ontology as well as the relational structure, a thorough analysis of the guidelines is required. This method creates a custom ontology structure and creates new relations that would be able to describe the processes in the guidelines in such a way that rule extraction is possible without keeping the original structure among guideline concepts. Automated ontology building from guideline models seems to be very realistic and has the potential of combining the practical data from guidelines with the capabilities of ontology models, simultaneously uniting several guidelines into one large shared structure.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por Fundamental Library of Latvia University of Life Sciences and Technologies