There are many problems requiring a semantic account of a database schema. At its best, such an account consists of mapping formulas between the schema and a formal conceptual model or ontology (CM) of the domain. This paper describes the underlying principles, algorithms, and a prototype of a tool which infers such semantic mappings when given simple correspondences from table columns in a relational schema to datatype properties of classes in an ontology. Although the algorithm presented is necessarily heuristic, we offer formal results stating that the answers returned are “correct” for relational schemas designed according to standard Entity-Relationship techniques. We also report on experience in using the tool with public domain schemas and ontologies.