Most databases are independently designed and, therefore, are usually implemented using different conceptual schemas, which creates a context of syntactic, structural and semantic-level heterogeneity. Nevertheless, when a set of databases refers to a common domain, it may become necessary to integrate them into a single database, or to intermediate access to the databases in a transparent way. To deal with the heterogeneity problem, it becomes necessary to align the conceptual schemas. This process is usually carried out by domain specialists, and tends to be tedious and error-prone. This dissertation presents the CatalogMatcher, a software infrastructure for catalog matching. A catalog stores data about a set of objects from a specific domain, typically classified by some sort of taxonomy or thesaurus. The CatalogMatcher contains components that implement instance-based alignment strategies.