AUTOMATION OF ONTOLOGY FORMATION FROM RELATIONAL DATABASE USING MAPPING RULES

• Main Author: Rendy
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42689
• Institution: Institut Teknologi Bandung

Ontology has an important role in representing the semantics of data sources. The development of information systems and database utilization in the era of big data makes ontology a need to represent the semantics of stored data. Building an ontology to represent knowledge about a specific domain from data sources is not an easy process. Several methods have been introduced to build ontologies from relational databases. However, there are limitations in building ontology with existing methods. In this thesis, we propose improvements on the existing mapping rules to enrich the semantics of ontology. We also propose a method to automatically obtain ontology from relational databases that are not in the third normal form (3NF). Normalization is carried out on the relational database scheme without changing the physical scheme of the relational database directly. Normalization is carried out based on denormalization reverse engineering. This is done because in general the existing operational database has a design with a normal scheme, but it is carried out denormalization to improve data processing performance. The results of the normalization scheme are mapped according to the mapping rules to form an ontology. We have developed a software to automatically form an ontology from relational database using the mapping rules. The mapping rules are implemented based on the implementation of existing mapping rules, the addition of new rules and by mapping components in the relational database into ontology components. The results of this study conclude that the construction of ontology from relational databases that are not in third normal form (3NF) can be done and the new mapping rules can enrich the rules of ontology mapping. Tests were also carried out on the results of the ontology formed by examining the syntax structure with Oops! and the results can be said to be very good.