Clustering and retrieval of web pages dominantly relies on analyzing either the content of individual web pages or the link structure between them. Some literature also suggests to use the structure of web pages, notably the structure of its DOM tree. However, little work considers the visual structure of web pages for clustering. In this paper (i) we motivate visual structure-based web page clustering and retrieval for a number of applications, (ii) we formalize a visual box model-based representation of web pages that supports new metrics of visual similarity, and (iii) we report on our current work on evaluating human perception of visual similarity of web pages and applying the learned visual similarity features to web page clustering and retrieval.

Traditionally, information extraction from web tables has focused on small, more or less homogeneous corpora, often based on assumptions about the use of tags. A multitude of different HTML implementations of web tables make these approaches difficult to scale. In this paper, we approach the problem of domain-independent information extraction from web tables by shifting our attention from the tree-based representation of web pages to a variation of the two-dimensional visual box model used by web browsers to display the information on the screen. The thereby obtained topological and style information allows us to fill the gap created by missing domain-specific knowledge about content and table templates. We believe that, in a future step, this approach can become the basis for a new way of large-scale knowledge acquisition from the current “Visual Web.”