Grant: Discovery Grants Program - Individual

Database integration is an increasingly important research area: Information systems development requires integrating existing systems and their databases; data warehousing requires the integration of data from multiple transaction processing systems for decision support; electronic business between organizations requires the integration of trade partners' business data for efficient business processes and communication; inter-agency information sharing in e-government requires integration of citizen data. Thus, improvements in database integration techniques have a wide range of applications and can benefit a variety of users. Database integration is a process with multiple steps, leading from the identification of the databases to the testing of the integrated system. The central step of this process is the identification of database elements that are similar to each other. This step is called schema matching. Human users perceive schema matching methods as successful when they match database elements according to the same similarity criteria that the users would apply. Many automatic software-supported schema matching methods have been proposed, using a range of different notions of similarity. However, existing research has paid little attention to empirical work: There is little knowledge of how human data integrators judge the similarity of database elements. A recent study has shown that a ""one-size-fits-all"" approach does not apply and that human notions of similarity are strongly affected by the database integration context. The proposed research is a study to systematically examine how human database integrators carry out schema matching. It will identify how context influences the schema matching process and outcomes. This knowledge will then be applied to implement a context-aware schema matching method. Context information is important to schema matching, but has been neglected by existing research. The use of context information in schema matching can have benefits to research and practice in areas such as information systems development, data warehousing, e-commerce, and e-government.