Abstract. We propose a novel approach to reverse engineering of relational databases to ontologies. Our approach is based on the idea that semantics of a relational database can be inferred, without an explicit analysis of relational schema, tuples and user queries. Rather, these semantics can be extracted by analyzing HTML forms, which are the most popular interface to communicate with relational databases for data entry and display on the Web. The semantics are supplemented with the relational schema and user “head knowledge” to build an ontology. Our approach can be applied to migrating data-intensive Web pages, which are usually based on relational databases, to the ontologybased Semantic Web.