In this report a framework for building “virtual” web sites using the HyperView system is presented. Virtual web sites are web sites that offer information extracted and integrated from other web sites on the fly. The HyperView system already supports the demand-driven integration of information from different semistructured information sources into a graph database. The problem we are dealing with here is to query the database and generate HTML pages from the results as a response to HTTP requests received from the user. The returned HTML pages should hide the aspects of data extraction and integration and should give the user the impression of a single, coherent web site. We show first how HyperViews comprised of graph-transformation rules can be defined that generate HTML pages from the database. This way web sites for individual application schemata can be designed. In the second part we present a generic rule set that defines a web interface for HyperView graph databases with arbitrary schemata. This generic web interface can be customized for the particular application by annotating the database schema and chosing appropriate styles. The work presented in this report completes the HyperView approach in the sense that it closes the circle of extracting and integrating information from the web by again publishing the integrated data on the web. Our approach applies as well to the integration and generation of XML documents on the web.