The access to Web data, its extraction, and its integration with data from other sources is a problem that can be addressed in different ways. Some approaches assume that the Web itself is a database, others aim to build a database view, materialized or not, of the Web. In this talk, we present an approach where (1) accesses to Web data are represented through an intermediate view mechanism (search views); where (2) implicit structure of the data (expressed within textual documents) is extracted thanks to textual extraction tools; and where (3) the cache is replaced by a database and is seen as the materialized database view of the Web built on the fly to answer a given query.