Database integration is an organization-level problem which has become more and more serious when correct and unified databases are required to support business intelligence and management decision making. In many organizations, information systems are developed independently. The database that supports the human resource management system, for example, is developed and implemented separately from the database that supports the main production system. Different naming conventions, data types, and values in such different databases make it hard to consider if the data items from different databases refer to the same real-world objects. The problems become worse in the case that databases have different data structures; different data models. Polyglot environments may be a solution for operational systems but may turn to be problems for decision support systems. This paper presents an approach to the database integration problem. Ontology is used as a central knowledge base where data items and relationships are identified and resolved. Since the database integration process must yield perfect or close to perfect result, any mismatches or errors are not acceptable and user involvements are required. Hence, a semi-automatic approach is adopted.
展开▼
