DEVELOPING AN ONTOLOGY SUPPORTED KNOWLEDGE BASE USING A RULE BASED SYSTEM

Razvoj semantičnega spleta in povezanih tehnologij je prišel do stopnje razvoja, kjer se sprašujemo, zakaj ne zasledimo več aplikacij na njegovi osnovi. Razlogi za to so prepogosto v zapletenih programskih vmesnikih in neizkušenih razvijalcih. V doktorski disertaciji naslavljamo omenjeno problematiko s formalno definicijo preslikave ontoloških komponent v objektno-orientirano programsko kodo. Verjamemo, da lahko s preslikavo olajšamo dostop in uporabo semantično opisanih podatkov v vsakodnevnih informacijskih sistemih. Preslikavo nadgrajujemo s sistemom na osnovi pravil, ki služijo kot osnova za izgradnjo baze znanja. Le-ta se od podatkov razlikuje v tem, da omogoča dinamično in ažurno prilagajanje glede na trenutno veljavne podatke in vnaprej pripravljena pravila. Splošni model smo aplicirali na dejanski ontološki in programski jezik. Prototipno implementacijo, imenovano SWROOLS, smo v raziskavi ovrednotili tako s stališča ohranjanja izraznosti pri preslikavi kot tudi učinkovitosti same baze znanja. Učinkovitost modela smo preverili s pomočjo eksperimenta in razširjene metodologije za vrednotenje tehnologij semantičnega spleta. V primerjavo smo vključili tri sorodne tehnologije iz nabora, ki ga ponuja semantični splet. Izkazalo se je, da je predstavljen model po učinkovitost vsaj enakovreden primerjanim semantičnim tehnologijam. Če upoštevamo omejitev implementacije preslikave, lahko zaključimo, da se je model izkazal kot superioren tako s stališča učinkovitosti kot tudi števila pridobljenih rezultatov.The development of the semantic web and its supporting technologies has gained such a level of maturity that the question arises, why don\u27t we see more applications being developed using semantic data. The reasons for this are mainly the complexity of ontology application interfaces and the inexperienced developers. These problems are addressed in this thesis using a formal procedure for mapping ontology artifacts to object-oriented code. We believe that using such a mapping, one can ease the development of everyday applications based on semantic data. The mapping is further enhanced by a rule based system, which is used to construct a knowledge base. Its main advantage in comparison to ordinary systems is its capability of automatic adaptation based on the current facts and a set of predefined rules. The general model is applied to a concrete ontology and object-oriented language. Its prototype implementation named SWROOLS was evaluated from the standpoint of lost expressivity by the mapping and performance characteristics of the knowledge base. The performance characteristics have been evaluated using an experiment and an extended benchmark for semantic web technologies. We compared our system with three well established approaches for building knowledge bases on the semantic web. It turned out that the introduced model has at least equivalent performance characteristics as the others. In fact, considering the deficits of the implemented mapping procedure, we can conclude that it is superior in terms of speed and the number of obtained results