Using Graph Transformation Techniques for Integrating Information from the WWW

The advent of the WWW has led to an abundance of information which is distributed on numerous Web sites of heterogeneous structure and coverage. It is therefore important to extract, combine and restructure these distributed data in order to facilitate the access to it. The HyperView methodology models Web sites and their HTML pages as graphs from which information is extracted by a hierarchy of views. We present the clustered graph data model (CGDM) used in the HyperView system. Views are defined using graph transformation rules. We use typed attributed Single Pushout graph transformation with application conditions on attributes. The main contribution of this paper is a new rule activation mechanism that supports the incremental computation of views, a crucial requirement in the context of information extraction from Web sites. The HyperView prototype is currently used in the field of Digital Libraries. In this paper, we demonstrate our methodology in the domain of town information.