A database conceptual schema is a high-level description of how database concepts are organized, which is typically as classes of objects and their attributes. A fundamental operation in many database applications, schema matching involves finding a mapping mu between the concepts in a source scheme S and the concepts in a target schema T such that, if t = mu(s), then s and t have the same meaning. Along with data warehousing, query mediation relies heavily on schema marching. This application uses a mediator to translate user queries, formulated in terms of a common schema M, into queries that local databases can handle. The mediator must therefore be able to match each local schema with M. Query mediation is particularly challenging in the context of the Web, where the number of local databases, over which the mediator has little control, is enormous. We examine three major approaches to schema matching - syntactic, semantic, and a priori - using examples, with a focus on mediator design.