Ontology approach for supporting context-aware B2B collaborations

• Main Author: Ang, Chin Teck
• Other Authors: Goh Eck Soong, Angela
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer applications::Physical sciences and engineering
• Online Access: http://hdl.handle.net/10356/16953
• Institution: Nanyang Technological University

Partner selection is a process that plays a pivotal role in the success of short-term B2B collaborations. A context-aware approach for partner selection was proposed and implemented as a XML-based system that selects and ranks potential partners based on a set of criteria. The XML-based system does not represent semantics of data well and can only compute new information needed for the ranking through an external rule engine with pre-defined B2B business rules to instruct the rule engine exactly what to do with the data. Interest was on building a system that can represent semantics of data and infer new data without the need to define external rules. OWL (Ontology Web Language) was identified as a possible alternative to XML to better represent semantics of data. One possible approach is to modify the XML-based system into an Ontology-based system that represents data as ontologies instead of XML profiles. The aim of this project is to investigate the possibility of incorporating the B2B business rules into ontology to reduce the need for external rules to be defined and compare the ranking performances of the Ontology-based system and the XML-based system. Through research and testing, it is concluded that the business rules defined for this project cannot be incorporated fully into ontology and have to be expressed in an OWL-based rule language. But the language is not powerful enough to express some of the business rules, this lead to inaccurate query results and subsequently affecting the ranking scores. The query evaluation was also identified as a performance bottleneck. Unlike the existing system, the new system does not need an external rule engine and the process of creating business rules is less tedious. The new system is also able to better express relationships between concepts and restrictions on these concepts. Overall, the XML-based system produced more accurate ranking results and outperforms the Ontology-based system in terms of execution time and memory usage. The existing system proved to be a more economical choice.