Schema Matching and Mapping

In Chapter 3 we described formalisms for specifying source descriptions and algorithms that use these descriptions to reformulate queries. To create the source descriptions, we often begin by creating semantic matches. The matches specify how the elements of the source schemas and the mediated schema semantically correspond to one another. Examples include “attribute name in one source corresponds to attribute title in another,” and “location is a concatenation of city, state, and zipcode.” In the next step we elaborate the matches into semantic mappings, which are typically structured queries written in a language such as SQL. The mappings specify how to translate data across the sources and the mediated schema. The above two steps are often referred to as schema matching and schema mapping, respectively, and are the focus of this chapter.