1 Some Practical Advice for Dealing with Semantic Heterogeneity in Federated.pdf

1Some Practical Advice for Dealing with Semantic Heterogeneity in Federated Database Systems Vincent Ventrone, The MITRE Corporation, Bedford, MA, vav@ Sandra Heiler, GTE Laboratories Incorporated, Waltham, MA, sh04@ Abstract For many organizations, database federation has become the preferred strategy for reconciling the proliferation of private and independent databases, and for increasing the productivity of their information technology investment. However, federation preserves the autonomy of the underlying systems, which are often based on different sets of assumptions. Hence, it does not directly address the problem of integrating semantically heterogeneous databases. The “integration” of component system data in the federation requires making database semantics, which are implicit in the component systems, explicit in metadata, and reconciling these differences in meaning. Currently, there is little technology available to “automate” this process. But, the problem cannot be avoided if organizations wish to provide semantically consistent access to multiple databases. Integrating semantically heterogeneous databases is costly, since it is a largely intellectual process involving extracting semantics, expressing them as metadata, and resolving inconsistencies among them for overlapping data. Thus, it is important to avoid expending the effort wherever possible. We suggest two ways to make this effort practical, based on our own experience with very large system integration efforts and discussions with others engaged in these efforts. First, narrow the scope of the problem and prioritize the information to be federated, based on requirements specifications for federation users. Second, exploit the various tools and enabling technologies that are available, including semantic and object models, Repository, CASE tools, and standards. 21. Introduction Over the past 15 years information systems in most organizations have been characterized by the rapid growth of per