The Deep Web comprises data stored in databases, dynamic pages, scripted pages and multimedia data, among other types of objects. The databases of the Deep Web are generally underrepresented by the search engines due to the technical challenges of locating, accessing and indexing them. The use of hyperlinks by search engines is not sufficient to achieve all the Deep Web data, requiring interaction with complex queries interfaces. This thesis presents an approach, called W-Ray, that provides visibility to Deep Web data. The approach relies on describing the relevant data through well-structured sentences, and on publishing the sentences as static Web pages. The sentences can be generated with embedded RDFa, keeping the semantics of the database. The Web pages thus generated are indexed by traditional Web crawlers and sophisticated crawlers that support semantic search. It is also presented a tool that supports the W-Ray approach. The approach has been successfully implemented for some real databases.