WF2OML: A Modeling Language for Mapping Web Forms to Ontology

Deep Web databases contain huge amounts of data that can only be accessed via the front-end Web form queries. Since query clients cannot directly issue SPARQL queries against the Deep Web databases, Semantic Web applications cannot take advantage of this information resource. Recently, many RDB2RDF mapping tools, relying on mapping languages such as W3C´s R2RML, provide the ability to view existing relational databases in the RDF model, and offer a virtual SPARQL endpoint over the mapped relational data. However, these mapping languages require the information provider to expose the underlying information structure, and to reconstruct the Web site. In this paper we propose a modeling language, WF2OML, for mapping Web forms to OWL or RDFS ontologies. Based on the WF2OML mapping, SPARQL endpoint queries against the back-end database of the Web form can be achieved. We introduce WF2OML, propose the architecture and algorithm design ideas of a WF2OML-based SPARQL query system, and demonstrate a mapping example using our implemented prototype. The experimental results on real-life Web form query sites (e.g., Amazon.com) confirm the feasibility of the proposed method.