Data integration involves merging two or more schemas into a global schema based on the identification of common concepts, and also the set of different concepts that are mutually related by some semantic properties. Identifying semantic relationships between a set of objects in one schema and a different set of objects in another schema is crucial to the integration process. Earlier integration strategies depend on significant manual input from the users and the ingenuity of the DBAs for identification and declaration of correspondence assertions. We argue that the task of manually identifying assertions for all attributes in the local databases is indeed enormous. In this paper, we design an integration algorithm that leverages a simple semistructured data model and a common ontology to dynamically identify correspondence assertions. We exploit set theory and logic to formally specify the assertions.